07_backtesting_with_zipline.ipynb missing train key in predictions.h5

nghnam · January 17, 2024, 2:56pm

I am running the notebook

PacktPublishing/Machine-Learning-for-Algorithmic-Trading-Second-Edition/blob/master/11_decision_trees_random_forests/07_backtesting_with_zipline.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Backtesting with Zipline - Using the Pipeline API with ML-driven Signals"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Imports & Settings"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {

This file has been truncated. show original

I ran file 05 and 06 to create predictions.h5, but can not find code which write train/1 to it

def load_predictions(bundle):
    t = 1
    df = pd.concat([pd.read_hdf(results_path / 'predictions.h5', 'train/{:02}'.format(t)), # missing train/1 key
                    pd.read_hdf(results_path / 'predictions.h5', 'test/{:02}'.format(t))])
    df = df[~df.index.duplicated()].drop('y_test', axis=1)
    predictions = df.iloc[:, :5].mean(1).to_frame('predictions')

    tickers = predictions.index.get_level_values('ticker').unique().tolist()

    assets = bundle.asset_finder.lookup_symbols(tickers, as_of_date=None)
    predicted_sids = pd.Int64Index([asset.sid for asset in assets])
    ticker_map = dict(zip(tickers, predicted_sids))

    return (predictions
            .unstack('ticker')
            .rename(columns=ticker_map)
            .predictions
            .tz_localize('UTC')), assets

Anyone can help ?

littlefrog · December 23, 2024, 3:34am

I got the same problem.

slothkins · February 16, 2025, 9:16am

same for me too…any luck?

Jiaqin-Tan · March 31, 2025, 1:01am

Can you read the HDF file, use:
readHDF = pd.HDFStore(file_name)

Then read the key:
key = readHDF.get(train/).values.tolist()

See what’s the output?