Time Series Analysis

Endpoint: POST /time-series-analysis

Request Parameters

File Upload

• file (required): The file containing the time series data to be processed. Supported formats include CSV, Excel, etc.

Time Column

• time_column (required): The name of the column containing the time-related data (e.g., "date", "timestamp").

Resampling Interval

• interval (optional): The frequency at which to resample the data. The options are:
  • "hourly"
  • "daily"
  • "weekly"
  • "monthly"
  • "yearly"
    Default is "daily".

Aggregation Functions

• aggregation_func (optional): Comma-separated list of aggregation functions to apply to the time series data. Default is ["mean"]. Available options: sum, mean, median, min, max, count, std.

Columns to Analyze

• columns_to_analyze (optional): A comma-separated list of columns to analyze. If omitted, all numeric columns will be analyzed.

Trend Tests

• trend_tests (optional): Comma-separated list of trend tests to apply. Default is ["adfuller"]. Available options: adfuller, kpss.

Decomposition

• decomposition (optional): Boolean indicating whether to decompose the time series data into trend, seasonal, and residual components. Default is false.

Decomposition Model

• decomposition_model (optional): The decomposition model to use. Available options: additive, multiplicative. Default is "additive".

Graphs

• include_graphs (optional): Boolean to specify if graphs should be included in the response. Default is false.
• graph_types (optional): Comma-separated list of graph types to include. Default is ["line"]. Available options: line, decomposition, acf, pacf, rolling_mean.

Example Request

curl -X POST "http://localhost:8000/time-series-analysis" \
-F "file=@data.csv" \
-F "time_column=date" \
-F "interval=daily" \
-F "aggregation_func=mean,median" \
-F "columns_to_analyze=Sales,Profit" \
-F "trend_tests=adfuller,kpss" \
-F "decomposition=true" \
-F "decomposition_model=additive" \
-F "include_graphs=true" \
-F "graph_types=line,decomposition"

Analysis Components

1. Aggregation: Aggregates the time series data based on the selected aggregation functions (e.g., mean, sum, median).
2. Trend Tests: Applies statistical tests like Augmented Dickey-Fuller (adfuller) and Kwiatkowski-Phillips-Schmidt-Shin (kpss) to detect trends in the data.
3. Decomposition: Decomposes the time series data into its trend, seasonal, and residual components.
4. Graphs: Optionally generates graphs such as line charts, decomposition components, autocorrelation function (ACF), partial autocorrelation function (PACF), and rolling mean.

Example Response

{
  "aggregated_data": {
    "mean": {"Sales": 5000, "Profit": 1200},
    "median": {"Sales": 5200, "Profit": 1100},
    "std": {"Sales": 800, "Profit": 200},
    "min": {"Sales": 1000, "Profit": 100},
    "max": {"Sales": 9000, "Profit": 2000}
  },
  "trend_tests": {
    "adfuller": {"statistic": -3.5, "pvalue": 0.01, "is_stationary": true},
    "kpss": {"statistic": 0.5, "pvalue": 0.1, "is_stationary": true}
  },
  "decomposition": {
    "trend": [100, 110, 120, 130],
    "seasonal": [20, 30, 25, 35],
    "residual": [5, 10, -5, 0]
  },
  "graphs": {
    "line": {
      "Sales": {"time": ["2021-01-01", "2021-01-02", "2021-01-03", "2021-01-04"], "values": [5000, 5200, 5100, 5300]},
      "Profit": {"time": ["2021-01-01", "2021-01-02", "2021-01-03", "2021-01-04"], "values": [1200, 1100, 1150, 1250]}
    },
    "decomposition": {
      "trend": {"time": ["2021-01-01", "2021-01-02", "2021-01-03", "2021-01-04"], "values": [100, 110, 120, 130]},
      "seasonal": {"time": ["2021-01-01", "2021-01-02", "2021-01-03", "2021-01-04"], "values": [20, 30, 25, 35]},
      "residual": {"time": ["2021-01-01", "2021-01-02", "2021-01-03", "2021-01-04"], "values": [5, 10, -5, 0]}
    }
  }
}