Pandas

Pandas is a fast, powerful, flexible, and easy-to-use data analysis and manipulation library built on top of NumPy. It provides DataFrames - tabular data structures perfect for handling experimental data.

Why Pandas?

• Intuitive: Work with labeled data using column names and row indices
• Flexible: Handle time series, categorical data, and mixed types
• Powerful: Built-in functions for grouping, pivoting, and merging
• I/O: Read/write many formats (CSV, Excel, SQL, HDF5, JSON)
• Integration: Works seamlessly with NumPy, Matplotlib, and other libraries

Key Features

DataFrames - Labeled Tables

import pandas as pd

# Create a DataFrame
data = pd.DataFrame({
    'subject': ['S01', 'S01', 'S02', 'S02'],
    'condition': ['A', 'B', 'A', 'B'],
    'reaction_time': [0.45, 0.52, 0.48, 0.55]
})

# Access columns
data['reaction_time']

# Filter rows
fast_trials = data[data['reaction_time'] < 0.5]

Data Manipulation

• Filtering: Select rows based on conditions
• Grouping: Aggregate by experimental conditions
• Merging: Combine multiple datasets
• Pivoting: Reshape data for analysis
• Missing Data: Handle NaN values intelligently

Time Series

Built-in support for temporal data:

# Create time index
dates = pd.date_range('2024-01-01', periods=100, freq='D')
data = pd.DataFrame({'value': range(100)}, index=dates)

# Resample to weekly averages
weekly = data.resample('W').mean()

Common Use Cases in Research

• Behavioral Data: Organize trial-by-trial responses
• Spike Counts: Tabulate firing rates across neurons and conditions
• Experimental Metadata: Track subject information, session details
• Statistical Analysis: Prepare data for statistical tests
• Exploratory Analysis: Quickly compute summary statistics

Getting Started

Install Pandas:

conda install pandas
# or
pip install pandas

Basic example:

import pandas as pd

# Load data
data = pd.read_csv('experiment_data.csv')

# Quick overview
print(data.head())
print(data.describe())

# Group by condition and compute means
results = data.groupby('condition')['reaction_time'].mean()

Tips

• Use .head() and .describe() to quickly inspect data
• Learn boolean indexing for powerful filtering
• Use .groupby() for aggregating across experimental conditions
• Set meaningful index names for easier data access
• Save intermediate results with .to_csv() or .to_hdf()
• Use method chaining for readable data pipelines