Data Scientist
Start here. Build forward.
Turn data into defensible findings, experiments and predictive baselines.
Follow the stages in order, or open a topic to check what you already know. Each panel includes learning content, a practice task and a role checkpoint. Save Learning, Done or Skipped as you go. Skipped topics do not count as completed.
Build your foundations
Python, files and debugging
Understand the concept
Python is the working language for many AI experiments. Learn to inspect small values before connecting a large model. Variables, lists, dictionaries, loops and functions are enough for a first useful script. Errors are information: read the traceback and isolate the smallest failing input.
Learn it step by step
- Create a function with explicit inputs and outputs.
- Read a small JSON file and validate required keys.
- Handle a missing file and malformed input.
- Write checks for normal and empty input.
Build a small proof
Build a script that counts tutorial topics from ten fictional records and reports invalid rows separately.
Ready to move on?
You can explain each line, reproduce one error and fix its cause without hiding the exception. Keep the expected output with the script.
Common mistake
Copying a large generated program without understanding its inputs makes even small changes difficult to verify.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
SQL and relational data
Understand the concept
SQL queries structured tables. Start with filtering, grouping and joins. A join can multiply rows when the relationship is not one-to-one, so check counts before trusting totals. Data engineers and scientists both need to understand what one row represents.
Learn it step by step
- Create fictional learners and enrolments tables.
- Select only necessary columns.
- Join using stable IDs.
- Compare row counts and inspect unmatched records.
Build a small proof
Calculate enrolment counts by course, including a course with zero learners.
Ready to move on?
Explain why an inner join may drop empty courses and how a left join changes that result. Verify totals against the input rows.
Common mistake
Joining on a non-unique display name can double-count or merge different people.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Probability, uncertainty and sampling
Understand the concept
Statistics helps you distinguish a pattern from noise. Learn averages, spread, distributions, sampling and confidence intervals before interpreting model scores. An estimate from a small or biased sample may not describe the wider population.
Learn it step by step
- Summarise a small synthetic dataset.
- Compare mean and median with an outlier.
- Draw repeated samples and observe changing estimates.
- Explain uncertainty in plain language.
Build a small proof
Simulate repeated coin tosses and plot how the estimated heads rate changes as the sample grows.
Ready to move on?
You can explain why two samples yield different estimates without concluding that one must be fraudulent or broken.
Common mistake
Correlation alone does not prove that changing one variable will cause another to change.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Learn the core skills
What is data cleaning?
In simple words
Imagine a registration list where Pune appears as “Pune”, “pune” and “ Pune”. Cleaning can standardise these values so your analysis counts them consistently. But cleaning is not deleting everything unusual. Some unusual entries are real, and some missing entries have a meaning you need to preserve.
Two useful terms
Missing value: an entry that is absent or unknown. Duplicate: a repeated record according to a defined identity rule.
Build your understanding
Cleaning transforms raw observations into consistent inputs. Keep the raw data and document your decisions so another person can reproduce the result. Correct obvious formatting issues, but do not silently erase unusual values that might represent real cases.
Follow the steps
- Standardise obvious whitespace and spelling variants in category names.
- Identify missing values separately from zero values.
- Detect duplicates using a meaningful record identifier.
- Define how to flag questionable records for review.
Worked example
Raw cities: [" Pune", "pune", "PUNE", "Mumbai"]
Normalised: ["pune", "pune", "pune", "mumbai"]
Record ID, not matching city alone, determines duplicates.Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Design a cleaning log with columns for row ID, original value, new value and reason. Apply it to five fictional rows.
Check your understanding
Two records from the same city are not duplicates merely because their city field matches. Duplicate rules depend on what one row represents.
A mistake to avoid
If cleaning uses learned statistics such as means, fit those statistics on training data only and apply them unchanged to validation and test data.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
What are training, validation and test data?
In simple words
When preparing for an exam, you study examples, use practice tests to improve and finally take an unseen exam. ML evaluation uses a similar separation. Training fits the model, validation guides your choices and testing checks the final design on held-out examples. The separation helps make your score more honest.
Two useful terms
Training set: examples used to fit a model. Held-out data: examples kept separate from fitting.
Build your understanding
A model needs evaluation on data it did not learn from. Training data fits parameters, validation data supports model choices and test data provides a final held-out check. The split must reflect how the system will be used: time, users and groups can create dependencies between rows.
Follow the steps
- Define what an independent example means for your task.
- Separate the test set before experimenting.
- Use validation data or cross-validation to choose settings.
- Evaluate the final design once on the held-out test set and report limitations.
Worked example
For 1,000 independent toy examples:
600 training + 200 validation + 200 test
For repeated users: keep each user in only one split.
For forecasting: use chronological splits.Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Explain why randomly splitting multiple messages from the same person can overestimate performance on entirely new people.
Check your understanding
The model may learn person-specific patterns shared across splits. A group-based split better tests generalisation to new people.
A mistake to avoid
Do not repeatedly tune choices against the final test score. That gradually turns the test set into another validation set.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
What is regression in machine learning?
In simple words
Regression predicts a number rather than a category. For example, a toy model might estimate a plant’s height from its age. Linear regression uses a line or a weighted combination of features. It describes a relationship in data, but a relationship alone does not prove that one thing causes another.
Two useful terms
Prediction: the estimated output. Residual: the difference between the observed and predicted value.
Build your understanding
Linear regression models a numeric target as a weighted combination of features plus an intercept. It is a useful baseline when relationships are approximately linear. A fitted line describes the training data; it does not establish causation or guarantee sensible extrapolation.
Follow the steps
- Create a small synthetic input and numeric target.
- Fit a linear regression model.
- Compare predictions with held-out observations.
- Inspect residuals, the difference between observed and predicted values.
Worked example
from sklearn.linear_model import LinearRegression
x = [[1], [2], [3], [4]]
y = [3, 5, 7, 9]
model = LinearRegression().fit(x, y)
print(model.predict([[5]])) # approximately [11.]Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Change one target from 9 to 20 and fit again. Describe how the unusual point changes predictions.
Check your understanding
In the original synthetic data, y = 2x + 1, so x = 5 gives 11. Real datasets rarely fit a line perfectly.
A mistake to avoid
Predicting far outside the observed range can be unreliable. Report the range of data used for fitting and examine outliers rather than automatically deleting them.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Practise with real decisions
What is classification in machine learning?
In simple words
Classification predicts a category. A message might be labelled spam or ordinary; a support request might be labelled billing or technical. Decide what each category means before collecting labels. A model cannot resolve inconsistent label rules just by seeing more examples.
Two useful terms
Class: one possible category. Classifier: a model that predicts a category.
Build your understanding
Classification assigns an input to a category. Start with a simple baseline and choose metrics that reflect your error costs. Accuracy alone can hide poor performance when one class dominates. Always inspect which cases the classifier gets wrong.
Follow the steps
- Define labels and ensure they are consistent.
- Measure class balance before fitting.
- Create a baseline that predicts the most common class.
- Compare a model using a confusion matrix and a suitable metric.
Worked example
Dataset: 95 ordinary messages, 5 spam messages
Always predict ordinary -> 95% accuracy
Spam recall -> 0 / 5 = 0
High accuracy, no useful spam detection.Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Propose two different costs: blocking a genuine message and missing a spam message. Explain how those costs influence your preferred threshold.
Check your understanding
Lowering a spam decision threshold usually catches more spam but may flag more genuine messages. Measure both effects on validation data.
A mistake to avoid
A predicted probability is not automatically calibrated. A score of 0.9 does not guarantee 90% correctness without checking calibration.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
What are precision and recall?
In simple words
Think about a lost-item search. Precision asks how many items you brought back were the right kind. Recall asks how many of all the right items you managed to find. A system can be cautious and precise while missing many valid items. Which error matters more depends on the task.
Two useful terms
False positive: an incorrect positive prediction. False negative: a positive case the system missed.
Build your understanding
Precision asks how many positive predictions were correct. Recall asks how many actual positives were found. They answer different questions, so neither should be interpreted alone. The right trade-off depends on the cost of false alarms and missed cases.
Follow the steps
- Build a confusion matrix with true and false positives.
- Calculate precision using predicted positives as the denominator.
- Calculate recall using actual positives as the denominator.
- Compare thresholds on validation examples before selecting one.
Worked example
True positives = 8, false positives = 2, false negatives = 4
Precision = 8 / (8 + 2) = 0.80
Recall = 8 / (8 + 4) = 0.667
F1 = 2 * precision * recall / (precision + recall) ≈ 0.727Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Imagine a search result list. Explain what adding many weak results could do to recall and precision.
Check your understanding
Adding results may find more relevant items and increase recall, while irrelevant additions reduce precision. The exact effect depends on which items are added.
A mistake to avoid
If a denominator is zero, choose and document a handling convention rather than silently reporting a misleading percentage.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
What is clustering?
In simple words
Clustering groups similar examples without requiring a label for every group beforehand. Imagine sorting unknown buttons by shape and size. Another person might sort the same buttons by colour and material. The groups depend on the features and similarity rules you choose, so there may not be one correct answer.
Two useful terms
Cluster: a group formed by the method. Distance: a measure of how different two examples are.
Build your understanding
Clustering groups examples according to a similarity definition. It does not discover objectively correct categories automatically. The chosen features, scaling and distance strongly affect the groups. Use clustering to explore patterns, then inspect whether they support a useful interpretation.
Follow the steps
- Choose numeric features for a synthetic collection.
- Scale features when their units differ substantially.
- Try a small range of cluster counts.
- Inspect representative examples and ambiguous boundary cases.
Worked example
from sklearn.cluster import KMeans
x = [[0, 0], [0, 1], [9, 9], [9, 8]]
m = KMeans(n_clusters=2, random_state=7, n_init=10).fit(x)
print(m.labels_)
# Group IDs are arbitrary; nearby points should share a group.Read the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Explain how using annual income in rupees alongside age in years could dominate an unscaled distance calculation.
Check your understanding
The feature with larger numeric differences may dominate distance even when it is not more meaningful. Scaling changes the geometry of the problem.
A mistake to avoid
Do not assign sensitive identities or make high-impact decisions about people from automatically generated clusters.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Ship, evaluate and improve
What is AI model evaluation?
In simple words
A polished answer is not enough to know whether an AI system works. Evaluation checks specific requirements using examples and scoring rules. For a course assistant, you might check whether it finds the correct course, admits missing information and respects account permissions. Keep failure cases as part of your test collection.
Two useful terms
Test case: an input with an expected behavior. Metric: a defined measurement of performance.
Build your understanding
An AI evaluation checks whether a system meets a defined task. Build examples around realistic successes and failures, not only easy demonstrations. Keep the test set stable enough to compare versions and record the model, prompt and settings used. A single aggregate score can hide critical mistakes.
Follow the steps
- Define success in observable terms.
- Create normal, ambiguous, missing-information and adversarial examples.
- Write expected answers or scoring rules before running the system.
- Compare versions and inspect regressions before accepting a change.
Worked example
FAQ evaluation columns:
question | expected source | required facts | forbidden claim | result
Example: fees unknown | FAQ-3 | say unknown | invented fee | pass/failRead the example once, then explain each part aloud in your own words. If it includes Python, you can understand the idea first and run the code later in a local Python environment. Code is shown as text; this page does not execute it.
Try this yourself
Write five tests for an enrolment assistant. Include a user without Pro access and a request for a nonexistent course.
Check your understanding
A response should not promise enrolment if server-side permissions deny it. Include permission-sensitive cases, not just wording checks.
A mistake to avoid
Do not judge a model solely by another model’s unreviewed score. Human checks and deterministic validation are useful for important requirements.
Before looking at the answer, write a short explanation and one example of your own. Finish by naming one limitation of the method. Use Mark as read when you are ready, or return to the previous lesson if a term is unclear.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Ablations and experimental controls
Understand the concept
An ablation removes or changes one component to test its contribution. Good experiments control other conditions and report variability where practical. If several things change together, it becomes hard to know why the result improved.
Learn it step by step
- Fix the evaluation set and baseline.
- Change one meaningful component.
- Repeat under comparable conditions.
- Report failures and uncertainty as well as wins.
Build a small proof
Compare a retrieval system with and without reranking on the same labelled questions.
Ready to move on?
You can attribute the comparison to one controlled difference and describe the limits of the small test set.
Common mistake
Selecting only the best run exaggerates performance. Document your selection method and all relevant runs.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Monitoring quality, latency and cost
Understand the concept
AI monitoring needs more than uptime. Track response time, failures, input volume and task quality using a suitable sample of reviewed outcomes. Quality can change when users or source documents change. Log enough to diagnose failures without retaining unnecessary sensitive input.
Learn it step by step
- Define a service-level target.
- Record request outcome and latency.
- Track a stable evaluation set over releases.
- Create an alert and an investigation checklist.
Build a small proof
Produce a small report comparing two releases on success rate, p95 latency and estimated per-request cost.
Ready to move on?
You can tell whether a change improved usefulness while increasing cost or response time, and explain the trade-off.
Common mistake
An average hides slow tails and rare severe failures. Examine distributions and critical error categories.
Your role checkpoint
For the Data Scientist path, connect this topic to your capstone: Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Build something you can explain.
Analyse a public or synthetic dataset, define a prediction task and publish a reproducible report with limitations.
Role expectations vary by company. Use this path as a starting point and build evidence of your skills.