Data Science with Python Cheatsheet

Data Science Cheatsheet (Beginner Friendly)

This document is written for absolute beginners. Every concept is explained in simple language, with context of why it is needed in real projects.

1. What is Data Science (Very Simple Explanation)

Data Science means using data to solve real business problems.

Companies collect huge amounts of data every day: customer details, purchases, clicks, reviews, sensor readings, etc. Raw data alone is useless. Data Science converts this raw data into useful information.

A Data Scientist:

  • Collects data

  • Cleans messy data

  • Finds patterns

  • Builds models

  • Makes predictions

  • Helps managers take decisions

Example business questions:

  • Which customers may leave?

  • Which product sells most?

  • What will be next month’s sales?

Python example:

print("Data Science = Data + Python + Statistics + Machine Learning")

2. Complete Real‑World Data Science Workflow

Every professional project follows these steps:

  1. Understand the business problem

  2. Collect data (CSV, database, API)

  3. Clean data (missing values, duplicates)

  4. Explore data (EDA)

  5. Create features

  6. Train model

  7. Evaluate model

  8. Deploy model

Beginners usually jump directly to ML. This is wrong. 70% of real work is cleaning + understanding data.

3. Python Basics for Data Science

Python is easy to read and widely used in Data Science.

Variables

age = 25
salary = 50000

Lists

numbers = [1, 2, 3, 4]

Loop

for n in numbers:
    print(n)

Function

def add(a, b):
    return a + b

print(add(3, 4))

Functions help reuse code.

4. NumPy (Numerical Calculations)

NumPy is used for fast mathematical operations.

Why NumPy:

  • Faster than Python lists

  • Used inside ML algorithms

import numpy as np
arr = np.array([10, 20, 30])
print(arr.mean())

5. Pandas (Working With Tables)

Pandas works like Excel but with programming.

import pandas as pd

df = pd.read_csv("data.csv")
print(df.head())

Common operations:

df["age"]
df[df["age"] > 30]

6. Handling Missing Values

Missing values confuse ML models.

Check missing:

df.isnull().sum()

Fill missing:

df["age"].fillna(df["age"].mean(), inplace=True)

Always understand why data is missing before filling.

7. Descriptive Statistics

Used to summarize data.

  • Mean = average

  • Median = middle

  • Mode = most frequent

print(df.describe())

Median is better for salary data because of outliers.

8. Exploratory Data Analysis (EDA)

EDA helps you understand patterns before ML.

You check:

  • Distribution

  • Relationships

  • Outliers

df.corr()

Never skip EDA.

9. Feature Engineering

Creating new useful columns from existing data.

df["income_per_age"] = df["income"] / df["age"]

Better features = better model.

10. Train Test Split

We never train on full data.

from sklearn.model_selection import train_test_split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

80% training
20% testing

11. Supervised Learning

Data has answers (labels).

Examples:

  • House price

  • Spam detection

from sklearn.linear_model import LinearRegression
model = LinearRegression()
model.fit(X_train, y_train)

12. Classification

Predict categories.

from sklearn.linear_model import LogisticRegression
model = LogisticRegression()
model.fit(X_train, y_train)

13. Prediction

pred = model.predict(X_test)

14. Model Evaluation

from sklearn.metrics import accuracy_score
accuracy_score(y_test, pred)

Accuracy alone is not enough for imbalanced data.

15. Confusion Matrix

Shows correct vs wrong predictions.

from sklearn.metrics import confusion_matrix
confusion_matrix(y_test, pred)

16. Overfitting vs Underfitting

Overfitting: model memorizes.
Underfitting: model too simple.

Solutions:

  • More data

  • Regularization

  • Better features

17. Cross Validation

More reliable evaluation.

from sklearn.model_selection import cross_val_score
cross_val_score(model, X, y, cv=5)

18. Feature Scaling

Important for distance-based models.

from sklearn.preprocessing import StandardScaler
scaler = StandardScaler()
X_scaled = scaler.fit_transform(X)

19. Unsupervised Learning

No labels.

from sklearn.cluster import KMeans
kmeans = KMeans(n_clusters=3)
kmeans.fit(X)

20. Pipelines (Professional Practice)

Prevents data leakage.

from sklearn.pipeline import Pipeline
pipe = Pipeline([
    ("scaler", StandardScaler()),
    ("model", LogisticRegression())
])

Final Beginner Advice

Learn in this order:

Python → Pandas → EDA → Statistics → Machine Learning → Projects

Build projects. Explain your thinking. Practice daily.

Scroll to Top