Time Series Analysis та аналіз часових рядів

Methods

ARIMA

Components: AR (autoregressive), I (integrated), MA (moving average).

Process: Stationarity check, parameter selection, fitting.

Application: For stationary series.

LSTM

Concept: Recurrent neural networks for temporal patterns.

Advantages: For nonlinear, complex patterns.

Application: For complex time series.

Concepts

Stationarity

Concept: Statistical properties do not change over time.

Tests: ADF test, KPSS test.

Achievement: Differencing, transformation.

Seasonality

Types: Daily, weekly, monthly, yearly.

Processing: Decomposition, seasonal adjustment.

Detection: ACF, PACF plots.

Trend

Types: Upward, downward, no trend.

Processing: Detrending, differencing.

Detection: Moving averages, regression.

Practical Examples

Example 1: ARIMA for forecasting

Stationarity: Check and achieve stationarity.

ARIMA: Determine the parameters (p, d, q).

Fitting: Train the ARIMA model.

Forecasting: Predict future values.

Example 2: LSTM for time series

Preparation: Prepare sequences for LSTM.

LSTM: Train the LSTM model.

Forecasting: Predict through LSTM.

Try it live

Everything above runs in your browser — open Dimensionality Reduction: PCA, t-SNE & UMAP and change the parameters while it is running. Nothing is installed, nothing is uploaded, the whole model lives in one tab.