Introduction to Data Science
An overview of core concepts, tools, and workflows in data science.
1. What Is Data Science?
Data science combines statistics, programming, and domain expertise to extract insights and build
predictive models from data. The typical workflow follows these stages:
1. Data Collection – gathering raw data (databases, APIs, files, sensors)
2. Data Cleaning – handling missing values, duplicates, and errors
3. Exploratory Data Analysis (EDA) – understanding patterns and relationships
4. Modeling – applying statistical or machine learning methods
5. Evaluation – measuring model performance
6. Communication – presenting insights via visualizations and reports
2. Key Statistical Concepts
Concept Description
Mean Average of values
Median Middle value when sorted
Mode Most frequent value
Standard Deviation Measure of spread around the mean
Correlation Strength of relationship between two variables (-1 to 1)
Distribution How values are spread (e.g., normal/bell curve)
3. Common Tools
Python with libraries:
pandas – data manipulation
numpy – numerical computing
matplotlib / seaborn – visualization
scikit-learn – machine learning
SQL – querying structured databases
Jupyter Notebooks – interactive coding and documentation
4. Working with Data (Pandas Example)
import pandas as pd
df = pd.read_csv("[Link]")
[Link]() # preview first rows
[Link]().sum() # check missing values
[Link](inplace=True) # remove missing rows
df["revenue"].mean() # compute average
[Link]("region")["sales"].sum() # aggregate by category
5. Data Visualization
Visualization helps identify patterns and communicate findings: - Bar chart – comparing
categories - Line chart – trends over time - Scatter plot – relationship between two variables -
Histogram – distribution of a single variable - Box plot – spread and outliers
import [Link] as plt
[Link](df["age"], bins=20)
[Link]("Age")
[Link]("Frequency")
[Link]()
6. Introduction to Machine Learning
Machine learning enables computers to learn patterns from data rather than being explicitly
programmed.
Types: - Supervised Learning: model learns from labeled data (e.g., predicting house prices
from features → regression; classifying emails as spam/not spam → classification) - Unsupervised
Learning: model finds patterns in unlabeled data (e.g., customer segmentation via clustering) -
Reinforcement Learning: model learns via trial and error with rewards (e.g., game-playing
agents)
Simple example — Linear Regression:
from sklearn.linear_model import LinearRegression
model = LinearRegression()
[Link](X_train, y_train) # train on features X, target y
predictions = [Link](X_test)
7. Model Evaluation
Task Common Metrics
Regression Mean Squared Error (MSE), R² score
Classification Accuracy, Precision, Recall, F1-score
Clustering Silhouette score
Overfitting occurs when a model performs well on training data but poorly on new data. Mitigate
with more data, simpler models, regularization, or cross-validation.
8. Practice Exercise
Given a dataset of student study hours and exam scores, outline the steps to build a model
predicting scores: 1. Load and clean the data (check for missing values/outliers). 2. Visualize the
relationship (scatter plot of hours vs. score). 3. Split data into training and test sets. 4. Train a
linear regression model. 5. Evaluate using MSE and R² on the test set.
Summary
Data science blends statistics, programming, and visualization to turn raw data into actionable
insight, with machine learning extending this into predictive and pattern-recognition capabilities.