Machine LearningEDABrowser-basedStudent demo

Interactive data analysis explorer

Univariate, Bivariate, and Multivariate Analysis Explorer

Understand how data analysis helps us study patterns before building machine learning models.

Before we build a machine learning model, we need to understand the data. Data is not just numbers in a table. Each column may contain useful information about people, objects, events, or behaviours.

Why do we study data patterns?

Machine learning models learn from patterns in data. If the data contains errors, missing values, strange values, weak relationships, or irrelevant variables, the model may perform poorly.

What does each variable look like?

Are there missing or unusual values?

Are two variables connected?

Which variables may help predict an outcome?

Are some variables repeating the same information?

Is the dataset suitable for machine learning?

Data analysis is the first step before machine learning because it helps us understand what the model will learn from.

1. Univariate Analysis

Univariate analysis means analysing one variable at a time.

If we analyse only the exam scores of students, we are doing univariate analysis.

Questions it helps answer

- What is the average exam score?
- What is the highest score?
- Are there missing values?
- Are there unusual values or outliers?

MeanMedianMinimumMaximumRangeStandard deviationFrequency countsHistogramBar chartBox plot

Univariate analysis is like looking at one column in a dataset and asking: "What does this variable look like by itself?"

2. Bivariate Analysis

Bivariate analysis means analysing two variables at the same time.

If we analyse study hours and exam scores together, we are doing bivariate analysis.

Questions it helps answer

- Do students who study more get higher scores?
- Is age related to income?
- Are two categories related?
- Is the relationship positive, negative, or weak?

Scatter plotCorrelationGroup comparisonCross-tabulationBox plot by categoryGrouped bar chartSimple linear regression

Bivariate analysis is like looking at two columns and asking: "Are these two variables connected?"

3. Multivariate Analysis

Multivariate analysis means analysing three or more variables at the same time.

If we analyse study hours, attendance, sleep hours, screen time, and exam score together, we are doing multivariate analysis.

Questions it helps answer

- Which variables are most important?
- Can we predict exam score using several variables?
- Are some variables strongly connected?
- Which variables should we use in a machine learning model?

Multiple regressionLogistic regressionCorrelation matrixHeatmapPCAClusteringDecision treesFeature importance

Multivariate analysis is like looking at many columns together and asking: "How do these variables work together?"

Comparison table

Type of Analysis	Number of Variables	Main Question	Example	Common Visualisation
Univariate	One variable	What does this variable look like?	Exam score only	Histogram, bar chart, box plot
Bivariate	Two variables	Are these two variables related?	Study hours vs exam score	Scatter plot, box plot, correlation
Multivariate	Three or more variables	How do many variables work together?	Study hours + attendance + sleep predicting exam score	Heatmap, regression, PCA, feature importance

Interactive Analysis Explorer

Explore a sample dataset or upload your own

Choose univariate, bivariate, or multivariate analysis. The tool detects column types, updates the selectors, calculates statistics, and explains what the results mean.

Dataset selector

Rows

Columns

Numerical columns

Categorical/text columns

Analysis type

Select one column

Univariate analysis: Exam Score

One variable by itself

Mean

64.5

Median

Minimum

Maximum

Range

Std. deviation

15.61

Missing values

Unique values

Histogram / distribution

40-482

48-562

56-640

64-722

72-802

80-882

The selected variable has an average value of 64.5. The lowest value is 40 and the highest value is 88. This helps us understand the overall distribution before comparing it with other variables.

Dataset preview

Sample student dataset - showing first 8 rows

Sample data

Student Name	Study Hours	Attendance Percentage	Sleep Hours	Screen Time	Exam Score	Pass/Fail
A	2	70	6	5	55	Pass
B	4	85	7	3	70	Pass
C	5	90	8	2	82	Pass
D	1	60	5	7	45	Fail
E	3	75	6	4	64	Pass
F	6	95	8	1	88	Pass
G	2	65	4	6	50	Fail
H	5	88	7	3	79	Pass

Student Name

categorical

Missing: 0 | Unique: 10

Study Hours

numerical

Missing: 0 | Unique: 6

Attendance Percentage

numerical

Missing: 0 | Unique: 10

Sleep Hours

numerical

Missing: 0 | Unique: 5

Screen Time

numerical

Missing: 0 | Unique: 8

Exam Score

numerical

Missing: 0 | Unique: 10

Pass/Fail

categorical

Missing: 0 | Unique: 2

Why is this important in machine learning?

Machine learning models learn patterns from data. If we do not understand the data first, we may build a poor model.

Understand the distribution of each variable

Find relationships between variables

Detect outliers and errors

Identify useful features

Remove irrelevant or duplicated variables

Understand possible bias in the dataset

Choose the right machine learning model

Explain results more clearly

For example, before predicting exam score, we should first check whether study hours, attendance, sleep, and screen time are related to the score.

Important: Correlation does not always mean causation

If two variables are related, it does not always mean one variable causes the other. Other factors such as attendance, previous knowledge, sleep, and motivation may also affect the result.

Start with one variable

Use univariate analysis first to understand each column separately. This helps detect missing values, unusual values, high variation, and general patterns.

Then compare two variables

Use bivariate analysis to check whether two variables are related. This helps identify useful predictors and possible relationships.

Finally study many variables together

Use multivariate analysis when the outcome depends on many factors. This is closer to real machine learning because models usually use multiple input features.

Check Your Understanding

Mini quiz

Score: 0/5

1. If we analyse only exam score, what type of analysis is it?

2. If we analyse study hours and exam score together, what type of analysis is it?

3. If we analyse study hours, attendance, sleep, and exam score together, what type of analysis is it?

4. Which visualisation is commonly used for two numerical variables?

5. Why is data analysis important before machine learning?

Summary

Univariate analysis studies one variable by itself. Bivariate analysis studies the relationship between two variables. Multivariate analysis studies three or more variables together.

These methods are important because machine learning depends on patterns in data. Before building a model, we should understand individual variables, relationships between variables, and the combined effect of multiple variables. A good machine learning workflow starts with data understanding before model building.