Time Series Analysis with Google Colab Training in Bangkok, Thailand

This course provides a hands-on introduction to time series analysis using Python in the Google Colab environment. Participants will learn the essential techniques for working with time-indexed data, including preprocessing, visualization, forecasting, and evaluation using libraries like Pandas, Statsmodels, Matplotlib, and Prophet.

Foundations of Time Series Analysis

Introduction to Time Series and Google Colab

• Overview of time series data
• Setting up Google Colab for time series work
• Loading datasets (CSV, APIs, Drive, etc.)
• Introduction to Pandas DateTime indexing

Time Series Data Preprocessing

• Converting columns to datetime objects
• Setting time indices and sorting
• Handling missing timestamps and duplicate values
• Frequency conversion (resampling, upsampling, downsampling)

Exploratory Time Series Visualization

• Line plots and time-indexed visualizations
• Plotting rolling means and standard deviation
• Visualizing seasonality and trends
• Introduction to autocorrelation (ACF) and partial autocorrelation (PACF)

Decomposition, Smoothing, and Forecasting Models

Time Series Decomposition

• Trend, seasonal, and residual components
• Additive vs multiplicative decomposition
• Using statsmodels.tsa.seasonal_decompose
• Visual interpretation of components

Smoothing Techniques and Stationarity

• Simple moving average (SMA)
• Exponential moving average (EMA)
• Differencing to remove trend/seasonality
• Augmented Dickey-Fuller (ADF) test for stationarity

Forecasting with ARIMA

• Understanding AR, MA, and ARIMA concepts
• Parameter selection (p, d, q)
• Auto ARIMA with pmdarima
• Building and evaluating ARIMA models
• Residual analysis and diagnostics

Advanced Forecasting and Real-World Projects

Forecasting with Facebook Prophet

• Introduction to Prophet for business forecasting
• Preparing data for Prophet (ds, y format)
• Adding holidays and seasonality
• Forecast visualization and components
• Cross-validation in Prophet

Evaluating Forecast Accuracy

• Train-test split in time series
• Performance metrics: MAE, MSE, RMSE, MAPE
• Visualizing forecast vs actual
• Tuning and improving model accuracy

Case Study and Dashboarding

• Build end-to-end time series workflow in Colab
• Combine charts, code, and markdown for presentation
• Tips for sharing and exporting notebooks

Final Exercises and Wrap-up

• Present forecasts with visual insights
• Peer review and instructor feedback
• Q&A and key takeaways

Final Assessment:

• Quiz on concepts and syntax
• Code challenge: Create a forecast and evaluate performance