Time Series Analysis MSFT
Time series forecasting project analyzing Microsoft stock prices using exploratory data analysis, statistical inspection, moving averages, and an LSTM neural network built with Keras.
Feb 2024 - Jun 2024 • 4 months
Tech Stack
PythonKerasLSTMPandasNumPyMatplotlibTime Series
Features
- Exploratory Data Analysis – Historical Microsoft stock price visualization and statistical inspection
- Time Series Decomposition – Trend, seasonal, and residual component extraction
- Moving Averages Analysis – Short-term and long-term smoothing for pattern detection
- LSTM Neural Network Model – Deep learning model trained on 60-day rolling windows
- Performance Evaluation – MSE, RMSE, and MAE metrics for train and test datasets
- Visual Comparison – Predicted vs real price comparison for model validation
- End-to-End Pipeline – Data preprocessing, normalization, training, evaluation, and visualization