Outliers Explained in Hindi l Machine Learning Course

1. Formula
   Observation

   Whiteboard text 'Marks: 15, 20, 25, 20, 100'.

Marks

Exam scores of five students out of 100.

Set of numerical values representing student scores.

1. Formula
   Observation

   Write on the whiteboard: "Marks: 15, 20, 25, 20, 100"

Marks: 15 , 20 , 25 , 20 , 100

A sample of 5 students' scores, used as an outlier example

A finite set of real numbers

1. Audio
   Observation

   The instructor repeatedly states that extra classes are needed only if the class average falls below 35

35

The average score threshold for determining whether a class needs extra lessons

A constant threshold on the score scale

1. Formula
   Observation

   The whiteboard displays ($15+20+25+20+100$)/5 = $180/5 = 36$

$\frac{15+20+25+20+100}{5}$

The arithmetic mean calculation for this sample

The sample mean formula

1. Formula
   Observation

   The whiteboard finally shows 36, which is circled and marked with a check

2. Audio
   Observation

   The instructor says the average goes up to 35

36

The average score result for this sample

real numbers

1. Diagram
   Observation

   The title at the top of the whiteboard reads Outliers.

Outliers

Outliers are observations that are extreme or unusual relative to most data points.

Data points in statistics and machine learning

1. Diagram
   Observation

   The whiteboard displays Marks: 15, 20, 25, 20, 100.

2. Audio
   Observation

   The instructor explains how an extreme value affects the overall average using this set of scores.

Marks: 15, 20, 25, 20, 100

A sample of student scores, where 100 is used as an example of an extreme value.

A discrete numerical sample

1. Formula
   Observation

   The whiteboard shows ($15+20+25+20+100$)/5 = $180/5 = 36$.

2. Audio
   Observation

   The instructor explains that this average influences the decision about whether to schedule an extra class.

($15+20+25+20+100$)/5 = $180/5 = 36$

Calculation of the arithmetic mean including the outlier 100.

Mean calculation

1. Audio
   Observation

   The instructor repeatedly says average falls below 35 or is less than 35.

2. Diagram
   Observation

   On the right side of the whiteboard, 35 is visible as a benchmark for comparison.

1. 35 appears in the video as a decision threshold, but its formal definition rule is not elaborated within this segment.

35

The average score threshold used to determine whether an extra class is needed.

The numerical threshold in the decision rule.

1. Audio
   Observation

   The instructor says, 'Instead of 100 here, let’s say it were 20.'

2. Diagram
   Observation

   On the whiteboard, the original 100 is changed to 20.

20

The new sample value after replacing the outlier 100, used to demonstrate the effect of changing a single extreme value.

Sample value replacement.

1. Formula
   Observation

   On the whiteboard, ($15+20+25+20+20$)/5 = $100/5 = 20$ is visible.

2. Audio
   Observation

   The instructor verbally calculates 100 divided by 5: 'How much does it come to?' It comes to be 20.

($15+20+25+20+20$)/5 = $100/5 = 20$

The arithmetic mean recalculated after removing the extreme value.

Mean calculation.

1. Diagram
   Observation

   In the lower right corner of the coordinate system, there is a dense cluster of small points.

2. Audio
   Observation

   The instructor calls this cluster 'majority of this data points'.

scatter cluster

A two-dimensional scatter cluster representing the majority of normal data points.

Scatter plot / regression illustration

1. Audio
   Observation

   The speaker defines outliers as unusual or extreme data points that deviate significantly from the typical value (majority of data points).

2. Diagram
   Observation

   The scatter plot shows a dense cluster of points and one isolated point far away.

Outliers are unusual or extreme data points that take very small or very large values compared to the majority of the dataset (typical value).

1. Applies to any dataset with records, observations, samples, or data points.

1. Formula
   Observation

   The title reads Outliers

2. Diagram
   Observation

   On the right coordinate plot, there is a cluster of dense points and one clearly separated high point

3. Audio
   Observation

   The instructor explains how the fraction 100 changes the class average

The video uses 'Outliers' as its title and illustrates through a sample of scores and a scatter plot that an outlier is a value in a data set that differs markedly from other observations and lies far from the main distribution. Here, 100 is the outlier relative to 15, 20, 25, and 20.

1. For intuitive understanding of outliers

2. The video does not provide a formal mathematical definition

1. Formula
   Observation

   Board work calculation: ($15 + 20 + 25 + 20 + 100$) / $5 = 180 / 5 = 36$

2. Audio
   Observation

   The instructor says 'just because of this person ... your average goes up to 35'

The method shown in the video uses the arithmetic mean to assess overall level, then explains how a single extreme value pulls the mean upward. Because the mean adds all observations equally and then divides by the count, a large outlier significantly affects the final result.

1. The sample consists of a finite set of numerical values

2. Using the arithmetic mean as the summary statistic

1. Intuitive introduction to outliers

1. Audio
   Observation

   The instructor says that most students need extra classes

2. Formula
   Observation

   In the sample, 15, 20, 25, and 20 are all below 35, while 100 is above 35

The video emphasizes a reasoning approach: if we consider the majority of individuals, then among 5 people, 4 have scores below 35, suggesting the entire class needs help; however, if we look only at the mean, adding a single value 100 shifts the mean to 36, reversing the conclusion. This shows that a single mean may obscure the internal structure of the sample.

1. When comparing the situation for the majority of individuals versus the overall mean

2. The threshold is set to 35

1. The mean is sensitive to outliers

1. Diagram
   Observation

   A dense cluster of small points and one isolated high point appear on the right axis

2. Animation
   Observation

   The instructor repeatedly points to this cluster and the isolated point

1. The axes are not labeled with specific variable names or units

2. The meaning of the left graph is not fully clear

The video uses a hand-drawn coordinate graph to illustrate outliers: most points cluster together, representing the main data distribution; one point lies far from this cluster, representing an anomalous observation. This diagram expresses "outlier" as a clear spatial separation.

1. For intuitive illustration only

2. Axes are not defined

1. Intuitive introduction to outliers

1. Audio
   Observation

   The instructor says, 'Such outliers have an impact and affect the result; this is an outlier.'

2. Diagram
   Observation

   The whiteboard title reads 'Outliers', with an outlier point marked far from the main cluster.

The video explains 'outlier' as 'an extreme or unusual value relative to most data points.' It is not defined via formal statistical tests, but illustrated by a point clearly separated from the main cluster: such a value strongly influences overall results.

1. There exists a group of data points that are mostly close to each other.

2. One observation appears extreme or unusual relative to these points.

1. Formula
   Observation

   The whiteboard sequentially shows ($15+20+25+20+100$)/5 = 36 and ($15+20+25+20+20$)/5 = 20.

2. Audio
   Observation

   The instructor emphasizes, 'I changed only one value, so it fell into the category of an extra class.'

The video demonstrates this by replacing only one value in the same set of scores: when 100 is replaced with 20, the mean drops from 36 to 20. This shows that the arithmetic mean fully incorporates extreme values into the overall level, so a single outlier can significantly alter the mean.

1. Using the arithmetic mean as a summary statistic

2. Small sample size

3. Presence of clear extreme observations

1. Intuitive definition of an outlier

1. Audio
   Observation

   The instructor says, 'And for this reason, the decision went against this class... now it is justified to hold an extra class for these students.'

2. Formula
   Observation

   Compare the positions of 36 and 20 relative to threshold 35.

1. The video does not provide the full formal statement of the extra class rule, only illustrating a decision example bounded by 35.

The video compares the mean against threshold 35: with the outlier, the mean is 36, which is above 35, so the instructor says the decision will oppose giving the whole class an extra class; after changing 100 to 20, the mean becomes 20, which is below 35, so the conclusion changes to these students needing an extra class. This example shows that outliers are not merely numerical deviations but may also alter downstream decisions.

1. A decision threshold based on the mean exists.

2. The threshold is set at 35.

3. The decision direction changes depending on whether the mean is high or low.

1. The mean is highly sensitive to outliers.

1. Diagram
   Observation

   The instructor draws a fitted line near the main cluster, then draws another line shifted toward the top-left outlier.

2. Audio
   Observation

   The instructor says, 'If your model is built, it will be built very well... but when outliers appear, it will try to tend, to move toward this point.'

1. The video does not write out a specific loss function or regression equation; it only uses geometric illustration to show how the trend line gets pulled off course.

The video uses a two-dimensional scatter plot to illustrate: if outliers are ignored, the model follows a reasonably sensible trend formed by most points; once a point far from the main cluster is included, the fitted line shifts toward that point, deviating from the original best optimal path. This is an intuitive explanation of how outliers affect model parameters.

1. Data can be represented as a two-dimensional scatter plot.

2. The model attempts to fit the overall trend.

3. There exist observations clearly distant from the main cluster.

1. Intuitive definition of an outlier

1. Formula
   Observation

   The top of the whiteboard displays the large heading 'Outliers'

The video screen displays the topic of 'outliers', but no specific definition is given in this 5-second segment.

1. Formula
   Observation

   The whiteboard shows the complete calculation for the average score: '($15+20+25+20+20$)/5 = $100/5 = 20$'

To calculate the mean, add up a set of numbers and then divide by the count of numbers. The video demonstrates this using scores as an example.

1. The data set contains a finite number of values.

1. Marks

1. Diagram
   Observation

   On the right side of the whiteboard, there is a scatter plot showing a line passing through the data points and an isolated point clearly deviating from that line.

In a scatter plot, a point far from both the other data points and the regression line is typically considered an outlier.

1. A two-dimensional coordinate system and a set of data points are present.

1. Formula
   Observation

   Board work computes 36

2. Audio
   Observation

   The instructor says average 35 ke upar chala jaata hai

For the sample 15, 20, 25, 20, 100, the arithmetic mean is 36, thus greater than the threshold 35.

1. The sample consists of the five numbers listed on the whiteboard

2. Arithmetic mean is used

3. The threshold is set to 35

This holds for the given sample

1. Audio
   Observation

   The instructor says four out of five children have scores below 35.

2. Formula
   Observation

   In the sample, 15, 20, 25, and 20 are all less than 35.

Among these five scores, four are below 35; therefore, based on the majority of individual observations, most students in this class may still need additional tutoring.

1. The threshold is set at 35.

2. Interpret 'majority' as four or more out of five observations.

This holds for the given sample.

1. Audio
   Observation

   The instructor says, 'just because of this person ... the school administration said no extra class will be held.'

2. Formula
   Observation

   The intuition shifts from 'most below 35' to $36 > 35$ using the mean.

1. The video does not explicitly state the mean after removing 100; it only states that most are below 35.

The video argues: solely because one person scored 100, the class average is pulled above 35, thereby reversing the decision about extra classes based on the mean.

1. The decision rule is 'hold extra classes only if the class average score is below 35'.

2. The sample contains an extreme high score 100.

This holds in the given example scenario.

1. Audio
   Observation

   The instructor says, 'The entire average was affected because of one child.'

2. Formula
   Observation

   Only one value is replaced in the same dataset, and the mean changes from 36 to 20.

In a small sample, a single extreme observation can significantly change the arithmetic mean.

1. Using the arithmetic mean

2. The sample contains obvious outliers.

Existence example: the video directly shows a set of 5 scores.

1. Audio
   Observation

   The instructor says 'jo decision hai woh is class ke khilaf gaya... ab in bacchon ka extra class lena jayaz hai.'

2. Formula
   Observation

   The relative sizes of 36 and 20 with respect to 35 differ.

1. The decision rule itself is given only at the example level, without general proof.

When the decision relies on comparing the mean to a threshold, outliers may reverse the decision conclusion.

1. The decision rule depends on the mean.

2. The threshold is fixed at 35.

3. The sample contains outliers.

Illustrative proposition, an informal general theorem.

1. Audio
   Observation

   The instructor says, 'I will try to go toward this point... slightly diverting from my best optimal path.'

2. Diagram
   Observation

   The directional difference between the two fitted lines visually demonstrates the shift.

1. The video does not provide formal definitions of the model class, loss function, or optimal path.

When observations far from the main cluster are present, the model's fitted line is pulled toward that point, thus deviating from the trend determined primarily by the majority of the data.

1. The data contain a main cluster and outliers.

2. The model attempts to fit the overall trend.

An intuitive illustrative proposition, demonstrated via scatter plots rather than rigorous proof.

1. Formula
   Observation

   The board work step-by-step writes ($15+20+25+20+100$)/5 = $180/5 = 36$.

2. Audio
   Observation

   The instructor verbally recites the summation and division by 5 in sync.

1. Expression

   $$15+20+25+20+100$$
   Explanation

   First, add up all five students' scores to obtain the total sum.

   Justification

   The definition of the arithmetic mean requires first computing the sum of all observations.

   Shown in the video
2. Expression

   $$\frac{15+20+25+20+100}{5}$$
   Explanation

   Divide the total sum by the number of observations 5, written as the mean expression.

   Justification

   The arithmetic mean equals the total sum divided by the sample size.

   Shown in the video
3. Expression

   $$\frac{180}{5}$$
   Explanation

   Compute the numerator to get 180.

   Justification

   $15+20+25+20+100=180$.

   Shown in the video
4. Expression

   $$36$$
   Explanation

   Continue dividing to obtain the average 36.

   Justification

   $180 \div 5 = 36$.

   Shown in the video

The arithmetic mean of this sample is 36, thus exceeding the threshold 35.

1. Audio
   Observation

   The presenter first says, 'majority-wise, an extra class is needed', then states, 'the average went above 35'.

2. Formula
   Observation

   Four numbers are below 35, yet the mean is 36.

1. The video omits the step 'mean after removing 100' and gives only a verbal judgment that the majority fall below the threshold.

1. Expression

   $$15,20,25,20 < 35$$
   Explanation

   First examine the majority of individuals and find that four out of five scores are below 35.

   Justification

   Directly compare each sample value with the threshold 35.

   Shown in the video
2. Expression

   $$\text{majority needs help}$$
   Explanation

   Based on this, the presenter draws the intuitive conclusion: most students in this class still need extra tutoring.

   Justification

   This is a verbal explanation from the video, not a formal theorem.

   Shown in the video
3. Expression

   $$\bar{x}=36>35$$
   Explanation

   Then compute the overall mean and find that, due to the presence of 100, the average rises to 36.

   Justification

   Use the arithmetic mean calculation completed earlier.

   Shown in the video
4. Expression

   $$\text{no extra class by rule}$$
   Explanation

   If the school decides solely based on "average score below 35", then this class is no longer classified as requiring extra classes.

   Justification

   Apply the decision rule given at the beginning of the video.

   Shown in the video

Under the same dataset, "looking at the majority of individuals" and "looking at the average" yield opposite management conclusions; this is precisely an example of how outliers affect the interpretability of the mean.

1. Formula
   Observation

   Two mean formulas are written on the whiteboard.

2. Audio
   Observation

   The instructor mentally computes $180/5=36$ and $100/5=20$ step by step.

1. Expression

   $$\frac{15+20+25+20+100}{5}=\frac{180}{5}=36$$
   Explanation

   First, add the five fractions to obtain the sum 180, then divide by the sample size 5 to get the mean 36 including the outlier.

   Justification

   Definition of arithmetic mean.

   Shown in the video
2. Expression

   $$\frac{15+20+25+20+20}{5}=\frac{100}{5}=20$$
   Explanation

   After replacing the original extreme value 100 with 20, the sum becomes 100; dividing by 5 yields the new mean 20.

   Justification

   Direct substitution of the same arithmetic mean formula into the new sample.

   Shown in the video
3. Expression

   $$36 \to 20$$
   Explanation

   Changing just one observation causes the mean to drop from 36 to 20.

   Justification

   Concluded from numerical comparisons in the previous two steps.

   Derived from the video

This derivation shows that the arithmetic mean is highly sensitive to a single extreme value.

1. Audio
   Observation

   The instructor compares 20 with 35 and, based on this, states that an extra class is needed.

2. Formula
   Observation

   35 remains on the whiteboard as the threshold.

1. The video does not fully state all conditions of the extra class rule.

1. Expression

   $$36 > 35$$
   Explanation

   The mean including the outlier exceeds the threshold 35, so, according to the video example, the decision does not support adding an extra class for the entire class.

   Justification

   Compare with the threshold.

   Shown in the video
2. Expression

   $$20 < 35$$
   Explanation

   The mean after replacing outliers is below threshold 35, so according to the video example, the decision shifts to supporting extra classes for these students.

   Justification

   Compare with the same threshold.

   Shown in the video

Under the same threshold, the presence or absence of outliers leads to opposite conclusions.

1. Audio
   Observation

   The speaker introduces an example of a class with 5 students and their exam marks out of 100.

2. Formula
   Observation

   The whiteboard lists 'Marks: 15, 20, 25, 20, 100'.

Identify outliers in a dataset of student exam marks.

1. The class has 5 students.

2. The exam is out of 100 marks.

3. The marks obtained are 15, 20, 25, 20, and 100.

Understand how outliers affect models using a relatable example.

1. Expression

   $$15, 20, 25, 20, 100$$
   Explanation

   List the marks of the five students.

   Justification

   Given in the video.

   Shown in the video

The mark 100 is an outlier compared to the other marks (15, 20, 25, and 20).

Visual inspection of the scatter plot shows 100 as an isolated point far from the cluster of the other marks.

1. Formula
   Observation

   Marks: 15 , 20 , 25 , 20 , 100

2. Audio
   Observation

   The presenter says "five students received their marks."

3. Formula
   Observation

   ($15+20+25+20+100$)/5 = $180/5 = 36$

Given the scores of 5 students in a class—15, 20, 25, 20, 100—and assuming the school mandates extra classes only if the class average falls below 35, determine whether this class requires additional instruction, and explain what this conclusion illustrates.

1. The sample is 15, 20, 25, 20, 100

2. The sample size is 5

3. The decision threshold is 35

4. The rule is to offer extra classes only if the average score is below 35.

Calculate the average score and determine the conclusion based on this rule; also understand how outliers affect the conclusion.

1. Expression

   $$\bar{x}=\frac{15+20+25+20+100}{5}$$
   Explanation

   Compute the class average using the arithmetic mean formula.

   Justification

   The video explicitly writes and uses the mean formula.

   Shown in the video
2. Expression

   $$\bar{x}=\frac{180}{5}=36$$
   Explanation

   Divide the sum by 5 to obtain the average score 36.

   Justification

   The board work shows the complete calculation.

   Shown in the video
3. Expression

   $$36>35$$
   Explanation

   Compare the average score with the threshold 35.

   Justification

   The video repeatedly uses 35 as the cutoff line.

   Shown in the video
4. Expression

   $$\text{No extra classes are required according to the rules}$$
   Explanation

   Since the average score is not below 35, the school will not add extra classes for this class under this rule.

   Justification

   Directly apply the administrative rule given in the video.

   Shown in the video
5. Expression

   $$\text{but four fifths of the students scored below thirty five}$$
   Explanation

   Meanwhile, most individual student scores remain below 35, indicating that the mean-based conclusion masks internal variation within the class.

   Justification

   This contrast is the video's central device for introducing the outlier problem.

   Shown in the video

The average score is 36; according to the rule "extra classes are added only if the average score is below 35", this class does not require extra classes; however, the video uses this example to illustrate how a single outlier 100 can raise the mean and thereby change the decision.

Verification can be done using the summation 180 and division $180/5=36$ shown on the board; alternatively, compare each of 15, 20, 25, 20 with 35 individually to confirm that most individuals fall below the threshold.

1. Diagram
   Observation

   The whiteboard lists Marks: 15, 20, 25, 20, 100 and their mean calculation.

2. Audio
   Observation

   The instructor uses the extra class example to explain how outliers affect decision-making.

1. The video does not specify the origin or scope of threshold 35.

Given a set of student scores, illustrate how one extreme high score affects the class mean and the judgment about assigning an extra class.

1. The score sample is 15, 20, 25, 20, 100

2. The decision threshold is 35

3. Replace 100 with 20 as a comparative scenario

Compare the mean before and after replacement, and explain why the decision changes.

1. Expression

   $$\frac{15+20+25+20+100}{5}=36$$
   Explanation

   First compute the mean of the original sample.

   Justification

   Definition of arithmetic mean.

   Shown in the video
2. Expression

   $$36 > 35$$
   Explanation

   The original mean exceeds the threshold, so the instructor states the decision will oppose adding an extra class for the entire class.

   Justification

   Compare with threshold 35.

   Shown in the video
3. Expression

   $$\frac{15+20+25+20+20}{5}=20$$
   Explanation

   Replace the outlier 100 with 20 and recalculate the mean.

   Justification

   Definition of arithmetic mean.

   Shown in the video
4. Expression

   $$20 < 35$$
   Explanation

   The new mean falls below the threshold, so the conclusion becomes that these students need an extra class.

   Justification

   Compare with threshold 35.

   Shown in the video

When the outlier is included, the mean is 36; after replacement, the mean is 20; since $36>35$ and $20<35$, the decision reverses.

Both mean calculations can be verified directly from the addition and division shown on the whiteboard.

1. Formula
   Observation

   The whiteboard displays specific score data and the calculation process.

Calculate the average of a set of student scores (15, 20, 25, 20, 20).

1. Score data: 15, 20, 25, 20, 20

Find the average of this data set.

1. Expression

   $$\frac{15+20+25+20+20}{5}$$
   Explanation

   Add all the scores and divide by the total number of scores.

   Justification

   Definition of the mean.

   Shown in the video
2. Expression

   $$\frac{100}{5}$$
   Explanation

   Compute the sum of the numerators

   Justification

   arithmetic addition

   Shown in the video
3. Expression

   $$20$$
   Explanation

   obtain the final average

   Justification

   arithmetic division

   Shown in the video

20

The calculation result on the whiteboard is circled and confirmed with a check mark.

1. Animation
   Observation

   The speaker draws a table, then plots its data points on a Cartesian coordinate system, showing a cluster and one isolated point.

1. Table

2. Cartesian axes

3. Data points (dots)

1. Drawing the table.

2. Drawing the axes.

3. Plotting the cluster of points.

4. Plotting the single outlier point.

1. The data comes from a single source or table.

Visualizes how an outlier appears as an extreme value separated from the main distribution of data points.

1. Diagram
   Observation

   The whiteboard reads 'Outliers' at the top, and below it 'Marks: 15 , 20 , 25 , 20 , 100'.

2. Diagram
   Observation

   On the right side, there is a coordinate axis, a cluster of densely packed points, and one isolated high point.

3. Diagram
   Observation

   On the left side, there is another rectangular or box-shaped sketch with a downward-pointing arrow; its meaning is unclear.

1. The variables for the left-side box plot and coordinate axis are not labeled.

2. Whether the isolated point exactly corresponds to 100 is not explicitly indicated by a number in the diagram.

1. Title: Outliers

2. List of scores: 15, 20, 25, 20, 100

3. Coordinate plot on the right side.

4. A cluster of densely packed points.

5. an isolated high point

6. left-side box sketch with a downward arrow

1. the instructor first points to the list of scores

2. then repeatedly points to the dense cluster and the isolated point on the right

3. finally writes the mean calculation formula on the whiteboard and circles and checks 36

1. the list of scores remains visible throughout

2. the contrast between the main cluster and the isolated point on the right remains visible throughout

the visual presentation grounds the abstract concept of an outlier at two levels: numerically, it is an extreme score 100; geometrically, it is a single point far from the main cluster in the scatter plot.

1. Animation
   Observation

   the instructor speaks while writing ($15+20+25+20+100$)/5

2. Formula
   Observation

   then writes =$180/5=36$ and circles and checks 36

1. the summation expression

2. denominator 5

3. result $180/5$

4. final result 36

1. first write the sum of the numerators

2. then write division by 5

3. finally write the result and emphasize it

1. the data used throughout are always the same set of five fractions

this animation transforms the verbal claim that "outliers affect conclusions" into verifiable numerical computation.

1. Diagram
   Observation

   At the start of the video, the title 'Outliers', the list of scores, the mean expression, and the scatter plot on the coordinate system all appear simultaneously.

1. Title: Outliers

2. List of scores: Marks: 15, 20, 25, 20, 100

3. Mean calculation

4. Two-dimensional coordinate system

5. Main cluster of data points

6. Upper-left outlier

1. The instructor points with a finger to score 100 and the outlier, establishing a correspondence between the numerical example and the graphical example.

1. The list of scores and the scatter plot both remain on the whiteboard, facilitating side-by-side understanding of how outliers affect the mean and how they affect the model.

The video first presents both aspects—'how outliers affect the average' and 'how outliers affect graphical trends'—on the same whiteboard.

1. Diagram
   Observation

   The instructor erases or covers the original 100 and writes 20 instead.

2. Formula
   Observation

   The mean calculation is rewritten as ($15+20+25+20+20$)/5 = $100/5 = 20$.

1. Original extreme value 100

2. New value 20

3. Modified mean expression

1. The last value in the sample changes from 100 to 20

2. The sum changes from 180 to 100

3. The mean changes from 36 to 20

1. The other four scores 15, 20, 25, 20 remain unchanged

2. The sample size remains 5

This visual modification emphasizes that changing just one value causes a large shift in the overall average.

1. Diagram
   Observation

   The instructor first draws a line through the main cluster, then draws another line shifted toward the upper-left outlier.

2. Audio
   Observation

   The instructor says the model would have been good, but after the outlier appears, it moves toward this point.

1. Two lines are not labeled with equations; they are only for illustration.

1. Main cluster of scatter points.

2. Outlier in the upper-left corner.

3. First fitted line.

4. Second fitted line, shifted.

1. A line that fits the main cluster more closely appears first.

2. Then a steeper line appears, shifted toward the outlier.

3. The instructor uses hand gestures to compare the directional differences between the two lines.

1. The positions of the scatter points themselves remain unchanged.

2. The outlier always lies outside the main cluster.

By contrasting two lines on the same scatter plot, the video intuitively demonstrates how outliers can alter the direction of the trend learned by the model.

1. Diagram
   Observation

   Throughout the segment, the instructor stands in front of a whiteboard that statically displays the title, data, calculation formulas, and charts.

1. Title 'Outliers'

2. Fractional data and the process of computing the average

3. Scatter plot and regression line

4. instructor

1. All text and graphics on the whiteboard remain unchanged.

The frame presents the conceptual background of outliers through static board work, including an example of mean calculation affected by an outlier and an intuitive scatter plot.

1. Audio
   Observation

   the presenter contrasts needing help for the majority versus not needing help after average 35 ke upar

seeing that the class average score exceeds a certain threshold, one assumes most students in the class do not need extra help.

the video uses 15,20,25,20,100 to illustrate: even if the average is 36, there are still 4 students below 35, so the mean cannot replace observing the individual distribution.

1. Audio
   Observation

   The presenter says just because of this person ... your average goes above 35

Thinking that an unusually high score will not noticeably change the overall statistical conclusion.

The video shows that a single outlier 100 is enough to push the mean above the threshold, thereby changing the decision on whether to hold extra classes.

1. Audio
   Observation

   The instructor repeatedly emphasized that only the value of ek was changed, so it fell under the consideration of the extra class.

2. Formula
   Observation

   The mean changed from 36 to 20.

Some might think that a single extreme value in the sample would have no substantial effect on the overall average or decision.

The video uses an example with 5 numbers to show that a single extreme value is enough to pull the mean from 36 to 20, and completely reverse judgments based on 35.

1. Audio
   Observation

   The instructor shifts from the mean example to discussing models and predictions.

2. Diagram
   Observation

   The fitted line is drawn shifted toward the outlier.

Outliers not only change the mean but also alter the direction of model fitting and prediction trends.

In the second half of the video, scatter plots and two fitted lines illustrate how an outlier pulls the model away from the best optimal path formed by the majority of the data.

1. Audio
   Observation

   The speaker explains that outliers are defined relative to the 'context value', which is the value of the majority of data points.

Outliers contrast with the context value (majority of data points) by being extremely small or large.

1. Audio
   Observation

   First explain outliers, then compute the mean affected by 100

2. Formula
   Observation

   The mean is computed directly using the sample that includes the outlier

The video first builds intuition using the concept of outliers, then demonstrates how outliers actually affect statistics through mean calculations.

1. Audio
   Observation

   The presenter directly links the average to the extra classes rule

In this example, "judging by the mean" and "judging by the majority of individuals" yield opposite conclusions; the video uses precisely this contrast to illustrate the outlier problem.

1. Diagram
   Observation

   The scatter plot on the right uses an isolated point to represent an observation far from the main cluster

1. The figure does not label numerical values directly

The scatter plot is a visual example of the outlier definition, using spatial separation to convey "anomalous observation".

1. Audio
   Observation

   The instructor first defines outlier, then immediately demonstrates its impact using the mean example.

2. Formula
   Observation

   The comparison between 36 and 20 follows immediately after the explanation of outliers.

The concept of outliers is introduced first, then immediately applied to an example illustrating the sensitivity of the arithmetic mean.

1. Audio
   Observation

   After comparing the average with 35, the instructor directly discusses the extra class decision.

2. Formula
   Observation

   The comparison between $36>35$ and $20<35$ forms the basis for the decision.

The change in the mean is further used to illustrate how outlier-driven threshold-based decisions can be reversed.

1. Audio
   Observation

   The instructor transitions from the effect on the average to the effect on models and predictions.

2. Diagram
   Observation

   The whiteboard shifts from numerical expressions to scatter plots and fitted lines.

The video generalizes the idea that outliers affect a single summary statistic to the idea that outliers also affect model-fitting trends.

1. Audio
   Observation

   The first half covers the extra class decision; the second half covers models and predictions.

2. Diagram
   Observation

   Numerical examples and graphical examples appear side by side in the same lesson.

Additional explanation: the former shows how outliers affect simple rule-based judgments; the latter shows how outliers affect the direction of continuous model fitting—these two levels differ but both demonstrate the consequences of outliers.

1. Diagram
   Observation

   The mean calculation and the scatter plot on the whiteboard are displayed side by side, suggesting a connection between them.

The video introduces the theme of how outliers affect statistics by showing the mean calculation for a data set containing extreme values such as 15 and 25.

1. Audio
   Observation

   The speaker explicitly asks 'What is an outlier?' and answers it.

1. Definition of Outliers

1. Audio
   Observation

   The speaker introduces the student marks example.

1. Student Exam Marks Example

1. Formula
   Observation

   Title: Outliers and the sample 15, 20, 25, 20, 100 appear together

1. Intuitive introduction to outliers
2. Illustration of an outlier in a scatter plot

1. Formula
   Observation

   ($15 + 20 + 25 + 20 + 100$) / $5 = 180 / 5 = 36$

1. The mean is sensitive to outliers
2. The calculation process of the sample mean.

1. Audio
   Observation

   The presenter compares the majority-wise result with the average result.

1. Conflict between the majority of observations and the mean
2. Most students fall below the threshold.

1. Audio
   Observation

   The extra classes rule and sample calculations run throughout this section.

1. An outlier example using five students' scores
2. After adding the outlier, the mean exceeds the threshold

1. Diagram
   Observation

   The coordinate graph on the right shows a cluster of points and one isolated point.

1. The axes are unlabeled.

1. Illustration of an outlier in a scatter plot
2. Whiteboard layout for teaching outliers.

1. Audio
   Observation

   The instructor explicitly says 'this is outlier'.

2. Diagram
   Observation

   The title and the isolated point jointly point to the concept.

1. Intuitive definition of an outlier

1. Formula
   Observation

   Two mean calculations are shown on the whiteboard.

2. Audio
   Observation

   The instructor emphasizes that only one value was changed.

1. The mean is highly sensitive to outliers.
2. Deriving the change in mean from the version with outliers to the version after replacement.

1. Audio
   Observation

   After comparing the mean with 35, the instructor discusses the extra class.

2. Formula
   Observation

   The contrast between $36>35$ and $20<35$ is visible.

1. Outliers cause decision reversal.
2. Comparing the two means using the threshold 35 leads to opposite decisions.

1. Diagram
   Observation

   The two fitted lines point in different directions.

2. Audio
   Observation

   The instructor says the model will move towards this point.

1. Outliers pull the regression/model fitting direction.
2. Animated chalkboard showing how the fitted line is pulled off course by an outlier.

1. Formula
   Observation

   The steps for computing the mean are clearly shown on the whiteboard.

1. calculating the mean
2. Calculate the average of student scores.