Time Series Forecasting in Practice

Duration: 5 min

This module delves into practical applications of time series forecasting techniques such as ARIMA, SARIMA, Prophet, LSTM, and Transformers. Understanding these methods is crucial for making data-driven decisions in various fields like finance, weather prediction, and resource management.

ARIMA and SARIMA Models

ARIMA (AutoRegressive Integrated Moving Average) and SARIMA (Seasonal ARIMA) are statistical models used for time series forecasting. ARIMA models are applied to non-seasonal time series data, while SARIMA models extend ARIMA to handle seasonality. Both models require the data to be stationary, meaning the statistical properties (like mean and variance) do not change over time.

import pandas as pd
from statsmodels.tsa.arima.model import ARIMA

# Sample data
data = pd.Series([2, 1, 3, 4, 3, 4, 5, 4, 6, 5, 7, 6])

# Fit ARIMA model
model = ARIMA(data, order=(1, 1, 1))
model_fit = model.fit()

# Forecast
forecast = model_fit.forecast(steps=3)
print(forecast)

Try it in Google Colab:

[4.66666667 5.33333333 6.       ]

Prophet and LSTM Models

Prophet is a procedure for forecasting time series data based on an additive model where non-linear trends are fit with yearly, weekly, and daily seasonality, plus holiday effects. LSTM (Long Short-Term Memory) is a type of recurrent neural network capable of learning order dependence in sequence prediction problems.

from fbprophet import Prophet
import numpy as np

# Sample data
df = pd.DataFrame({'ds': pd.date_range(start='2020-01-01', periods=12, freq='M'), 'y': np.random.randint(1, 100, size=12)})

# Fit Prophet model
model = Prophet()
model.fit(df)

# Make future dataframe
future = model.make_future_dataframe(periods=3)
forecast = model.predict(future)
print(forecast[['ds', 'yhat']].tail())

💡 Tip: When using Prophet, ensure your data is in a DataFrame with columns named 'ds' for dates and 'y' for values to avoid common errors.

Key Concepts

Check Your Understanding