Data Science Methodology Practice Flashcards

Comprehensive vocabulary flashcards covering the Data Science Methodology, evaluation metrics, and analytic types based on the Unit 2 Class XII AI curriculum.

Data Science Methodology

A process with a prescribed sequence of iterative steps that data scientists follow to approach a problem and find a solution.

John B. Rollins

The Data Scientist from IBM Analytics who put forward the 10-step framework for solving data science problems systematically.

Business Understanding

Step 1 of the methodology where the data scientist understands the problem by asking questions, identifying customer needs, and defining objectives.

5W1H Problem Canvas

A tool used during the Business Understanding stage to deeply understand the issue by answering Who, What, Where, When, Why, and How.

Analytic Approach

Step 2 of the methodology where the data scientist defines how to use data to answer the question, choosing between descriptive, diagnostic, predictive, or prescriptive techniques.

Descriptive Analytics

A type of analytics that answers "What happened?" using graphs, charts, and summary statistics like mean, median, and mode.

Diagnostic Analytics

A type of analytics that answers "Why did it happen?" using root cause analysis, hypothesis testing, and correlation.

Predictive Analytics

A type of analytics that answers "What will happen?" using regression, classification, or clustering to forecast future outcomes.

Prescriptive Analytics

A type of analytics that answers "What should we do?" using optimization, simulation, and decision analysis to recommend actions.

Data Requirements

Step 3 of the methodology where the data scientist identifies the types of data required (numbers, words, images), their structure, and sources.

Structured Data

Data organized in tables, rows, and columns, such as customer databases or spreadsheets.

Unstructured Data

Data with no predefined structure, such as social media posts, images, and videos.

Semi-Structured Data

Data that has some organization but is not fully tabular, such as emails, XML files, and JSON.

Data Collection

Step 4 of the methodology involving the systematic process of gathering observations or measurements from primary or secondary sources.

Primary Data Source

Original firsthand data collected through surveys, interviews, experiments, or IoT sensors.

Secondary Data Source

Pre-existing data from sources like books, journals, and online databases such as data.gov, World Bank, Kaggle, or WHO.

Data Understanding

Step 5 of the methodology where data scientists use descriptive statistics and visualization to check if collected data represents the problem.

Data Preparation

Step 6 and the most time-consuming stage, involving cleaning data, combining sources, and transforming data into meaningful input variables.

Feature Engineering

The process of selecting, modifying, or creating new features from raw data to improve model performance during Data Preparation.

AI Modelling

Step 7 where the prepared dataset is used to build descriptive or predictive models based on the defined analytical approach.

Model Evaluation

Step 8 where the model is checked to determine if it answers the original question satisfactorily before deployment.

Train-Test Split

A validation method where the dataset is divided into a training set (commonly $80\%$) and a testing set (commonly $20\%$).

K-Fold Cross Validation

A validation method where data is divided into $k$ equal folds; the model trains on $k-1$ folds and tests on the remaining fold, repeating the process $k$ times.

Deployment

Step 9 of the methodology involving rolling out the model into a production environment for use by stakeholders.

Feedback

Step 10 of the methodology where performance data is collected from users to improve the model in future iterations.

Confusion Matrix

A 2x2 table used to evaluate the performance of a binary classification model by showing True Positives, True Negatives, False Positives, and False Negatives.

True Positive (TP)

A result in a confusion matrix where the model predicted YES and the actual value was YES.

True Negative (TN)

A result in a confusion matrix where the model predicted NO and the actual value was NO.

False Positive (FP)

A Type I Error where the model predicted YES but the actual value was NO.

False Negative (FN)

A Type II Error where the model predicted NO but the actual value was YES.

Precision

The metric measuring the accuracy of positive predictions, calculated as $\frac{TP}{TP + FP}$.

Recall

The metric measuring how many actual positives were correctly identified, calculated as $\frac{TP}{TP + FN}$.

F1 Score

The harmonic mean of Precision and Recall, calculated as $2 \times \frac{\text{Precision} \times \text{Recall}}{\text{Precision} + \text{Recall}}$.

Accuracy

The metric for overall correct predictions out of all predictions, calculated as $\frac{TP + TN}{TP + FP + FN + TN}$.

Mean Absolute Error (MAE)

The sum of absolute differences between predictions and actual values, calculated as $\frac{\sum | \text{Actual} - \text{Predicted} |}{n}$.

Mean Squared Error (MSE)

The mean of squared distances between predicted and actual values, calculated as $\frac{\sum (\text{Actual} - \text{Predicted})^2}{n}$.

Root Mean Squared Error (RMSE)

The square root of MSE, calculated as $\sqrt{MSE}$, which is preferred because it is in the same units as the target variable.

BAD-UP-MEDF

A mnemonic for the 10 steps: Business Understanding, Analytic Approach, Data Requirements (and Collection), Data Understanding, Data Preparation, AI Modelling, Evaluation, Deployment, and Feedback.