القائمة
Generate HDF5 data files from tushare/extra_data for a given index (CSI300/CSI1000/etc). Produces daily_pv_{index}.h5 with 58 raw fields, used by build_features_from_h5.py and fin_factor.
التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.
استنادا إلى تصنيف SOC المهني
CSI300 训练+预测流水线。使用 daily_pv_csi300.h5，提取 Alpha158(158)+42 个增强因子=200 个特征，执行 walk-forward LightGBM 训练和预测。
Automated factor mining pipeline using rdagent fin_factor + DeepSeek LLM. Discovers, validates, and records non-duplicate quantitative factors across diverse categories (momentum, value, quality, growth, volatility, liquidity, leverage, cash flow, size, dividend).
CSI1000 训练+预测流水线。使用 daily_pv_csi1000.h5，默认提取 csi_1000.md(7 个独立有效因子)+Alpha158(158) = 165 特征；可选 practice_factor.md(42 个实践因子)+Alpha158(158) = 200 特征。执行 walk-forward LightGBM 训练和预测。
| name | gen_tushare_h5 |
| description | Generate HDF5 data files from tushare/extra_data for a given index (CSI300/CSI1000/etc). Produces daily_pv_{index}.h5 with 58 raw fields, used by build_features_from_h5.py and fin_factor. |
Reads tushare/extra_data/{SYMBOL}/*.csv, filters by index constituents, extracts 58 raw fields, and outputs an HDF5 file for downstream factor computation.
# CSI300 (default)
python3 scripts/generate_h5.py
# CSI1000
python3 scripts/generate_h5.py csi1000
python3 scripts/generate_h5.py csi1000 --output /tmp/daily_pv_csi1000.h5
# CSI500 / CSI800 / all stocks
python3 scripts/generate_h5.py csi500
python3 scripts/generate_h5.py all
rdagent_workspace/factor_data_template/daily_pv_{index}.h5
MultiIndex DataFrame: (datetime, instrument) × 58 columns (market + valuation + fundamental + financial).
| File | Role |
|---|---|
scripts/generate_h5.py | Main extraction script (parallel, calendar building, merge) |
tushare/cn_data/instruments/{index}.txt | Index constituent definitions (read-only) |
tushare/extra_data/{SYMBOL}/*.csv | Raw TuShare CSV data (read-only) |
python3 -c "
import pandas as pd
h5 = pd.HDFStore('rdagent_workspace/factor_data_template/daily_pv_csi300.h5', 'r')
print(f'Shape: {h5.get_storer(\"data\").shape}')
cols = h5.get_storer('data').non_index_axes[0][1]
print(f'Columns: {len(cols)}')
insts = h5.get_storer('data').non_index_axes[1][1]
print(f'Instruments: {len(insts)}')
h5.close()
"