Time Series Forecasting with ARIMA and Pandas

Write a Python program to build a time series forecasting model using ARIMA and Pandas.

The task involves building a time series forecasting model using the ARIMA (AutoRegressive Integrated Moving Average) technique and the Pandas library for data manipulation. The process includes generating or obtaining a time series dataset, splitting it into training and testing sets, and then fitting an ARIMA model to the training data. The model is used to make predictions on the test data, and its performance is evaluated using metrics such as the mean squared error. Visualization of the original and forecasted data helps in understanding the model's accuracy and performance.

Sample Solution:

Python Code :

# Import necessary libraries
# Import pandas for data manipulation and analysis
import pandas as pd
# Import numpy for numerical operations
import numpy as np
# Import matplotlib for plotting
import matplotlib.pyplot as plt
# Import ARIMA from statsmodels for time series forecasting
from statsmodels.tsa.arima.model import ARIMA
# Import mean_squared_error from sklearn for model evaluation
from sklearn.metrics import mean_squared_error

# Generate a sample time series dataset
def generate_sample_data():
    # Create a date range
    date_range = pd.date_range(start='2020-01-01', periods=100, freq='D')
    # Generate random data
    data = np.random.rand(100) * 100
    # Create a DataFrame
    df = pd.DataFrame(data, index=date_range, columns=['value'])
    # Return the DataFrame
    return df

# Function to plot the time series data
def plot_time_series(data, title='Time Series Data'):
    # Set the plot size
    plt.figure(figsize=(10, 6))
    # Plot the data
    plt.plot(data, label='Original Data')
    # Set the plot title
    # Set the x-axis label
    # Set the y-axis label
    # Display the legend
    # Show the plot

# Function to split the dataset into training and testing sets
def train_test_split(data, split_ratio=0.8):
    # Calculate the split point
    split_point = int(len(data) * split_ratio)
    # Split the data into training and testing sets
    train, test = data[:split_point], data[split_point:]
    # Return the training and testing sets
    return train, test

# Function to fit ARIMA model and make predictions
def arima_forecast(train, test, order=(5,1,0)):
    # Initialize the history with the training data
    history = [x for x in train]
    # Initialize the list of predictions
    predictions = []
    # Iterate over the testing data
    for t in range(len(test)):
        # Fit the ARIMA model
        model = ARIMA(history, order=order)
        model_fit = model.fit()
        # Make a one-step forecast
        output = model_fit.forecast()
        # Get the forecast value
        yhat = output[0]
        # Append the forecast value to the predictions list
        # Append the actual value to the history
    # Return the list of predictions
    return predictions

# Function to evaluate the model
def evaluate_forecast(test, predictions):
    # Calculate the mean squared error
    error = mean_squared_error(test, predictions)
    # Print the mean squared error
    print(f'Test Mean Squared Error: {error:.3f}')
    # Set the plot size
    plt.figure(figsize=(10, 6))
    # Plot the actual data
    plt.plot(test.index, test, label='Actual Data')
    # Plot the predicted data
    plt.plot(test.index, predictions, color='red', label='Predicted Data')
    # Set the plot title
    plt.title('Actual vs Predicted')
    # Set the x-axis label
    # Set the y-axis label
    # Display the legend
    # Show the plot

# Main function
def main():
    # Generate sample data
    df = generate_sample_data()
    # Plot the original time series data
    plot_time_series(df, title='Original Time Series Data')
    # Split the data into training and testing sets
    train, test = train_test_split(df['value'])
    # Fit ARIMA model and make predictions
    predictions = arima_forecast(train, test)
    # Evaluate the forecast
    evaluate_forecast(test, predictions)

# Run the main function
if __name__ == '__main__':


time series forecasting model using ARIMA and Pandas.png


  • Import necessary modules: "random" for generating random numbers and "numpy" for numerical operations.
  • Define the Rastrigin function to optimize.
  • Set genetic algorithm parameters: population size, genome length, crossover rate, mutation rate, number of generations, and bounds for gene values.
  • Define a function to initialize the population with random individuals within the given bounds.
  • Define a function to evaluate the fitness of each individual in the population using the Rastrigin function.
  • Implement tournament selection to select individuals based on their fitness.
  • Implement single-point crossover to create new individuals by combining parts of parent chromosomes.
  • Implement mutation to introduce random changes to individuals with a certain probability.
  • Define the main genetic algorithm function:
  • Initialize the population.
  • Iterate over the number of generations:
  • Evaluate fitness of the population.
  • Select individuals based on fitness.
  • Perform crossover to create the next generation.
  • Apply mutation to the next generation.
  • Re-evaluate fitness of the new population.
  • Track and print the best fitness of each generation.
  • Return the best individual found.
  • Uncomment and run the genetic algorithm to find and print the best solution.

Python Code Editor :

Have another way to solve this solution? Contribute your code (and comments) through Disqus.

Previous: Python Library for Polynomial Arithmetic.
Next: Monitor and Alert System Resource usage with Python.

What is the difficulty level of this exercise?

Test your Programming skills with w3resource's quiz.

Follow us on Facebook and Twitter for latest update.