Duplicate from derek-thomas/probabilistic-forecast

.gitattributes

@@ -0,0 +1,34 @@
+*.7z filter=lfs diff=lfs merge=lfs -text
+*.arrow filter=lfs diff=lfs merge=lfs -text
+*.bin filter=lfs diff=lfs merge=lfs -text
+*.bz2 filter=lfs diff=lfs merge=lfs -text
+*.ckpt filter=lfs diff=lfs merge=lfs -text
+*.ftz filter=lfs diff=lfs merge=lfs -text
+*.gz filter=lfs diff=lfs merge=lfs -text
+*.h5 filter=lfs diff=lfs merge=lfs -text
+*.joblib filter=lfs diff=lfs merge=lfs -text
+*.lfs.* filter=lfs diff=lfs merge=lfs -text
+*.mlmodel filter=lfs diff=lfs merge=lfs -text
+*.model filter=lfs diff=lfs merge=lfs -text
+*.msgpack filter=lfs diff=lfs merge=lfs -text
+*.npy filter=lfs diff=lfs merge=lfs -text
+*.npz filter=lfs diff=lfs merge=lfs -text
+*.onnx filter=lfs diff=lfs merge=lfs -text
+*.ot filter=lfs diff=lfs merge=lfs -text
+*.parquet filter=lfs diff=lfs merge=lfs -text
+*.pb filter=lfs diff=lfs merge=lfs -text
+*.pickle filter=lfs diff=lfs merge=lfs -text
+*.pkl filter=lfs diff=lfs merge=lfs -text
+*.pt filter=lfs diff=lfs merge=lfs -text
+*.pth filter=lfs diff=lfs merge=lfs -text
+*.rar filter=lfs diff=lfs merge=lfs -text
+*.safetensors filter=lfs diff=lfs merge=lfs -text
+saved_model/**/* filter=lfs diff=lfs merge=lfs -text
+*.tar.* filter=lfs diff=lfs merge=lfs -text
+*.tflite filter=lfs diff=lfs merge=lfs -text
+*.tgz filter=lfs diff=lfs merge=lfs -text
+*.wasm filter=lfs diff=lfs merge=lfs -text
+*.xz filter=lfs diff=lfs merge=lfs -text
+*.zip filter=lfs diff=lfs merge=lfs -text
+*.zst filter=lfs diff=lfs merge=lfs -text
+*tfevents* filter=lfs diff=lfs merge=lfs -text

.gitignore

@@ -0,0 +1,2 @@
+.idea
+lightning_logs

AirPassengers.csv

@@ -0,0 +1 @@
+Month,#Passengers

README.md

@@ -0,0 +1,14 @@
+---
+title: Probablistic Forecasting
+emoji: 🐨
+colorFrom: red
+colorTo: gray
+sdk: gradio
+sdk_version: 3.27.0
+app_file: app.py
+pinned: false
+license: apache-2.0
+duplicated_from: derek-thomas/probabilistic-forecast
+---
+
+Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference

app.py

@@ -0,0 +1,74 @@
+import gradio as gr
+import pandas as pd
+from gluonts.dataset.pandas import PandasDataset
+from gluonts.dataset.split import split
+from gluonts.torch.model.deepar import DeepAREstimator
+
+from make_plot import plot_forecast, plot_train_test
+
+
+def offset_calculation(prediction_length, rolling_windows, length):
+    row_offset = -1 * prediction_length * rolling_windows
+    if abs(row_offset) > 0.95 * length:
+        raise gr.Error("Reduce prediction_length * rolling_windows")
+    return row_offset
+
+
+def preprocess(input_data, prediction_length, rolling_windows, progress=gr.Progress(track_tqdm=True)):
+    df = pd.read_csv(input_data.name, index_col=0, parse_dates=True)
+    row_offset = offset_calculation(prediction_length, rolling_windows, len(df))
+    return plot_train_test(df.iloc[:row_offset], df.iloc[row_offset:])
+
+
+def train_and_forecast(input_data, prediction_length, rolling_windows, epochs, progress=gr.Progress(track_tqdm=True)):
+    if not input_data:
+        raise gr.Error("Upload a file with the Upload button")
+    try:
+        df = pd.read_csv(input_data.name, index_col=0, parse_dates=True)
+    except AttributeError:
+        raise gr.Error("Upload a file with the Upload button")
+
+    row_offset = offset_calculation(prediction_length, rolling_windows, len(df))
+
+    gluon_df = PandasDataset(df, target=df.columns[0])
+
+    training_data, test_gen = split(gluon_df, offset=row_offset)
+
+    model = DeepAREstimator(
+            prediction_length=prediction_length,
+            freq=gluon_df.freq,
+            trainer_kwargs=dict(max_epochs=epochs),
+            ).train(
+            training_data=training_data,
+            )
+
+    test_data = test_gen.generate_instances(prediction_length=prediction_length, windows=rolling_windows)
+    forecasts = list(model.predict(test_data.input))
+    return plot_forecast(df, forecasts)
+
+
+with gr.Blocks() as demo:
+    gr.Markdown("""
+    # How to use
+    Upload a univariate csv with the first column showing your dates and the second column having your data
+    
+    # How it works
+    1. Click **Upload** to upload your data
+    2. Click **Run**
+        - This app will visualize your data and then train an estimator and show its predictions 
+    """)
+    with gr.Accordion(label='Hyperparameters'):
+        with gr.Row():
+            prediction_length = gr.Number(value=12, label='Prediction Length', precision=0)
+            windows = gr.Number(value=3, label='Number of Windows', precision=0)
+            epochs = gr.Number(value=10, label='Number of Epochs', precision=0)
+    with gr.Row():
+        upload_btn = gr.UploadButton(label="Upload")
+        train_btn = gr.Button(label="Train and Forecast")
+    plot = gr.Plot()
+
+    upload_btn.upload(fn=preprocess, inputs=[upload_btn, prediction_length, windows], outputs=plot)
+    train_btn.click(fn=train_and_forecast, inputs=[upload_btn, prediction_length, epochs, windows], outputs=plot)
+
+if __name__ == "__main__":
+    demo.queue().launch()

make_plot.py

@@ -0,0 +1,114 @@
+from typing import List
+
+import numpy as np
+import pandas as pd
+import plotly.graph_objects as go
+
+
+def plot_train_test(df1: pd.DataFrame, df2: pd.DataFrame) -> go.Figure:
+    """
+    Plot the training and test datasets using Plotly.
+
+    Args:
+        df1 (pd.DataFrame): Train dataset
+        df2 (pd.DataFrame): Test dataset
+
+    Returns:
+        None
+    """
+
+    # Create a Plotly figure
+    fig = go.Figure()
+
+    # Add the first scatter plot with steelblue color
+    fig.add_trace(go.Scatter(
+            x=df1.index,
+            y=df1.iloc[:, 0],
+            mode='lines',
+            name='Training Data',
+            line=dict(color='steelblue'),
+            marker=dict(color='steelblue')
+            ))
+
+    # Add the second scatter plot with yellow color
+    fig.add_trace(go.Scatter(
+            x=df2.index,
+            y=df2.iloc[:, 0],
+            mode='lines',
+            name='Test Data',
+            line=dict(color='gold'),
+            marker=dict(color='gold')
+            ))
+
+    # Customize the layout
+    fig.update_layout(
+            title='Univariate Time Series',
+            xaxis=dict(title='Date'),
+            yaxis=dict(title='Value'),
+            showlegend=True,
+            template='plotly_white'
+            )
+    return fig
+
+
+def plot_forecast(df: pd.DataFrame, forecasts: List[pd.DataFrame]) -> go.Figure:
+    """
+    Plot the true values and forecasts using Plotly.
+
+    Args:
+        df (pd.DataFrame): DataFrame with the true values. Assumed to have an index and columns.
+        forecasts (List[pd.DataFrame]): List of DataFrames containing the forecasts.
+
+    Returns:
+        go.Figure: Plotly figure object.
+    """
+
+    # Create a Plotly figure
+    fig = go.Figure()
+
+    # Add the true values trace
+    fig.add_trace(go.Scatter(
+            x=pd.to_datetime(df.index),
+            y=df.iloc[:, 0],
+            mode='lines',
+            name='True values',
+            line=dict(color='black')
+            ))
+
+    # Add the forecast traces
+    colors = ["green", "blue", "purple"]
+    for i, forecast in enumerate(forecasts):
+        color = colors[i]
+        for sample in forecast.samples:
+            fig.add_trace(go.Scatter(
+                    x=forecast.index.to_timestamp(),
+                    y=sample,
+                    mode='lines',
+                    opacity=0.15,  # Adjust opacity to control visibility of individual samples
+                    name=f'Forecast {i + 1}',
+                    showlegend=False,  # Hide the individual forecast series from the legend
+                    hoverinfo='none',  # Disable hover information for the forecast series
+                    line=dict(color=color)
+                    ))
+        # Add the average
+        mean_forecast = np.mean(forecast.samples, axis=0)
+        fig.add_trace(go.Scatter(
+                x=forecast.index.to_timestamp(),
+                y=mean_forecast,
+                mode='lines',
+                name=f'Mean Forecast',
+                line=dict(color='red', dash='dash')
+                ))
+
+    # Customize the layout
+    fig.update_layout(
+            title='Passenger Forecast',
+            xaxis=dict(title='Index'),
+            yaxis=dict(title='Passenger Count'),
+            showlegend=True,
+            legend=dict(x=0, y=1, font=dict(size=16)),
+            hovermode='x'  # Enable x-axis hover for better interactivity
+            )
+
+    # Return the figure
+    return fig

requirements.txt

@@ -0,0 +1,3 @@
+gluonts[torch,pro]
+pandas
+plotly