Logistic Regression in Python Getting Data – Data Collection and Preparation Guide

Logistic Regression in Python – Getting Data

Data is the foundation of every machine learning project. Even the most advanced algorithms cannot produce meaningful results without quality data. Before training a Logistic Regression model, the first step is to collect, load, and understand the dataset that will be used for classification.

In this tutorial, you will learn how to obtain data for Logistic Regression projects, load datasets into Python, inspect data quality, and prepare data for machine learning tasks.

By the end of this guide, you will understand how to build a reliable data pipeline for Logistic Regression applications.

---

Why Data Matters

Machine learning models learn patterns from data.

The quality of predictions depends heavily on:

• Data accuracy
• Data completeness
• Data consistency
• Data relevance
• Data quantity

Poor-quality data often leads to poor model performance.

A common saying in data science is:

Garbage In, Garbage Out (GIGO)

This means that inaccurate or incomplete data produces unreliable predictions.

---

Types of Data Used in Logistic Regression

Logistic Regression is primarily used for classification problems.

Examples include:

Application	Target Variable
Email Filtering	Spam / Not Spam
Medical Diagnosis	Sick / Healthy
Customer Churn	Leave / Stay
Loan Approval	Approved / Rejected
Product Purchase	Buy / Not Buy

The target variable typically contains categorical values represented as:

0 = Negative Class
1 = Positive Class

---

Sources of Machine Learning Data

Data can come from various sources.

CSV Files

One of the most common formats.

Example:

Age,Salary,Purchased
25,30000,0
35,65000,1
45,85000,1

Advantages:

• Easy to create
• Human-readable
• Supported by Pandas

---

Excel Files

Many businesses store information in Excel spreadsheets.

Example:

customers.xlsx

Useful for:

• Sales reports
• Financial records
• Customer databases

---

Databases

Large applications often store data in:

• MySQL
• PostgreSQL
• SQLite
• SQL Server

Example:

SELECT * FROM customers;

---

APIs

Data can be retrieved from web services.

Examples:

• Weather APIs
• Finance APIs
• Social media APIs
• E-commerce APIs

---

Public Datasets

Popular sources include:

• Kaggle
• UCI Machine Learning Repository
• Government Open Data Portals
• Academic Research Datasets

These datasets are excellent for practice and learning.

---

Creating a Sample Dataset

For this tutorial, we will use customer purchase data.

Create a file named:

customers.csv

Contents:

Age,Salary,Purchased
22,25000,0
25,30000,0
30,45000,0
35,60000,1
40,70000,1
45,85000,1
50,95000,1

Save the file inside:

data/customers.csv

---

Installing Pandas

Pandas is the most popular Python library for data analysis.

Install it using:

pip install pandas

Verify installation:

import pandas as pd

print(pd.__version__)

---

Loading Data from a CSV File

Use Pandas to load the dataset.

import pandas as pd

data = pd.read_csv(
    "data/customers.csv"
)

print(data)

Output:

   Age  Salary  Purchased
0   22   25000          0
1   25   30000          0
2   30   45000          0
3   35   60000          1

The dataset is now available as a DataFrame.

---

Viewing the First Rows

Inspect the first few records.

print(data.head())

Output:

   Age  Salary  Purchased
0   22   25000          0
1   25   30000          0
2   30   45000          0
3   35   60000          1
4   40   70000          1

This helps verify that data was loaded correctly.

---

Viewing the Last Rows

Display the last records.

print(data.tail())

Output:

   Age  Salary  Purchased
2   30   45000          0
3   35   60000          1
4   40   70000          1
5   45   85000          1
6   50   95000          1

---

Checking Dataset Information

Use the info() method.

print(data.info())

Example output:

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 7 entries
Data columns (total 3 columns):

This shows:

• Number of rows
• Number of columns
• Data types
• Missing values

---

Understanding Data Types

Check column types.

print(data.dtypes)

Output:

Age          int64
Salary       int64
Purchased    int64

Machine learning models require numerical values, so data types should be verified before training.

---

Statistical Summary

Generate summary statistics.

print(data.describe())

Output:

             Age       Salary
count   7.000000     7.000000
mean   35.285714  58571.428571

Useful metrics include:

• Mean
• Minimum
• Maximum
• Standard deviation

---

Checking Missing Values

Missing values can negatively impact model performance.

Check for missing data:

print(data.isnull().sum())

Output:

Age          0
Salary       0
Purchased    0

A value greater than zero indicates missing data.

---

Handling Missing Values

Example dataset:

Age  Salary  Purchased
25   30000   0
NaN  45000   1
35   NaN     1

Remove missing values:

data = data.dropna()

Or replace them:

data.fillna(
    data.mean(),
    inplace=True
)

---

Removing Duplicate Records

Check duplicates.

print(data.duplicated().sum())

Remove duplicates.

data = data.drop_duplicates()

Duplicate records can distort model training.

---

Selecting Features and Target

Separate input variables and labels.

X = data[['Age', 'Salary']]

y = data['Purchased']

Features:

Age
Salary

Target:

Purchased

This structure is required by Scikit-Learn.

---

Loading Data from Excel

Install support package:

pip install openpyxl

Read Excel files:

data = pd.read_excel(
    "customers.xlsx"
)

This is useful when working with business spreadsheets.

---

Loading Data from SQL Databases

Example using SQLite:

import sqlite3
import pandas as pd

conn = sqlite3.connect(
    "customers.db"
)

data = pd.read_sql_query(
    "SELECT * FROM customers",
    conn
)

Databases are common in production systems.

---

Saving Cleaned Data

After cleaning the dataset:

data.to_csv(
    "data/cleaned_customers.csv",
    index=False
)

Benefits:

• Faster future loading
• Consistent preprocessing
• Reproducible results

---

Best Practices for Data Collection

Use Reliable Sources

Ensure data accuracy and credibility.

Gather Relevant Features

Collect variables related to the target prediction.

Maintain Data Quality

Remove errors and inconsistencies.

Keep Data Organized

Use dedicated folders for datasets.

Document Data Sources

Track where each dataset originated.

Store Raw Data Separately

Keep original data untouched.

---

Common Beginner Mistakes

Avoid these mistakes:

• Ignoring missing values
• Using duplicate records
• Mixing categorical and numerical data incorrectly
• Forgetting to inspect datasets
• Training models before cleaning data
• Using irrelevant features

---

Example Workflow

A typical machine learning data workflow:

1. Collect Data
2. Load Data
3. Inspect Data
4. Clean Data
5. Handle Missing Values
6. Remove Duplicates
7. Select Features
8. Train Model

Following this process improves model reliability.

---

Conclusion

Getting data is the first and most important step in building a Logistic Regression model. High-quality data allows machine learning algorithms to learn meaningful patterns and produce accurate predictions.

In this tutorial, you learned how to obtain data from different sources, load datasets into Python, inspect data quality, handle missing values, remove duplicates, and prepare features for machine learning. With a clean and organized dataset, you are now ready to move on to data preprocessing, feature engineering, and Logistic Regression model training using Scikit-Learn.