|
| 1 | +# Copyright (c) Microsoft Corporation. |
| 2 | +# Licensed under the MIT License. |
| 3 | + |
| 4 | +import numpy as np |
| 5 | +import pandas as pd |
| 6 | +import lightgbm as lgb |
| 7 | + |
| 8 | +from qlib.model.base import ModelFT |
| 9 | +from qlib.data.dataset import DatasetH |
| 10 | +from qlib.data.dataset.handler import DataHandlerLP |
| 11 | +import warnings |
| 12 | + |
| 13 | + |
| 14 | +class HFLGBModel(ModelFT): |
| 15 | + """LightGBM Model for high frequency prediction""" |
| 16 | + |
| 17 | + def __init__(self, loss="mse", **kwargs): |
| 18 | + if loss not in {"mse", "binary"}: |
| 19 | + raise NotImplementedError |
| 20 | + self.params = {"objective": loss, "verbosity": -1} |
| 21 | + self.params.update(kwargs) |
| 22 | + self.model = None |
| 23 | + |
| 24 | + def _cal_signal_metrics(self, y_test, l_cut, r_cut): |
| 25 | + """ |
| 26 | + Calcaute the signal metrics by daily level |
| 27 | + """ |
| 28 | + up_pre, down_pre = [], [] |
| 29 | + up_alpha_ll, down_alpha_ll = [], [] |
| 30 | + for date in y_test.index.get_level_values(0).unique(): |
| 31 | + df_res = y_test.loc[date].sort_values("pred") |
| 32 | + if int(l_cut * len(df_res)) < 10: |
| 33 | + warnings.warn("Warning: threhold is too low or instruments number is not enough") |
| 34 | + continue |
| 35 | + top = df_res.iloc[: int(l_cut * len(df_res))] |
| 36 | + bottom = df_res.iloc[int(r_cut * len(df_res)) :] |
| 37 | + |
| 38 | + down_precision = len(top[top[top.columns[0]] < 0]) / (len(top)) |
| 39 | + up_precision = len(bottom[bottom[top.columns[0]] > 0]) / (len(bottom)) |
| 40 | + |
| 41 | + down_alpha = top[top.columns[0]].mean() |
| 42 | + up_alpha = bottom[bottom.columns[0]].mean() |
| 43 | + |
| 44 | + up_pre.append(up_precision) |
| 45 | + down_pre.append(down_precision) |
| 46 | + up_alpha_ll.append(up_alpha) |
| 47 | + down_alpha_ll.append(down_alpha) |
| 48 | + |
| 49 | + return ( |
| 50 | + np.array(up_pre).mean(), |
| 51 | + np.array(down_pre).mean(), |
| 52 | + np.array(up_alpha_ll).mean(), |
| 53 | + np.array(down_alpha_ll).mean(), |
| 54 | + ) |
| 55 | + |
| 56 | + def hf_signal_test(self, dataset: DatasetH, threhold=0.2): |
| 57 | + """ |
| 58 | + Test the sigal in high frequency test set |
| 59 | + """ |
| 60 | + if self.model == None: |
| 61 | + raise ValueError("Model hasn't been trained yet") |
| 62 | + df_test = dataset.prepare("test", col_set=["feature", "label"], data_key=DataHandlerLP.DK_I) |
| 63 | + df_test.dropna(inplace=True) |
| 64 | + x_test, y_test = df_test["feature"], df_test["label"] |
| 65 | + # Convert label into alpha |
| 66 | + y_test[y_test.columns[0]] = y_test[y_test.columns[0]] - y_test[y_test.columns[0]].mean(level=0) |
| 67 | + |
| 68 | + res = pd.Series(self.model.predict(x_test.values), index=x_test.index) |
| 69 | + y_test["pred"] = res |
| 70 | + |
| 71 | + up_p, down_p, up_a, down_a = self._cal_signal_metrics(y_test, threhold, 1 - threhold) |
| 72 | + print("===============================") |
| 73 | + print("High frequency signal test") |
| 74 | + print("===============================") |
| 75 | + print("Test set precision: ") |
| 76 | + print("Positive precision: {}, Negative precision: {}".format(up_p, down_p)) |
| 77 | + print("Test Alpha Average in test set: ") |
| 78 | + print("Positive average alpha: {}, Negative average alpha: {}".format(up_a, down_a)) |
| 79 | + |
| 80 | + def _prepare_data(self, dataset: DatasetH): |
| 81 | + df_train, df_valid = dataset.prepare( |
| 82 | + ["train", "valid"], col_set=["feature", "label"], data_key=DataHandlerLP.DK_L |
| 83 | + ) |
| 84 | + |
| 85 | + x_train, y_train = df_train["feature"], df_train["label"] |
| 86 | + x_valid, y_valid = df_train["feature"], df_valid["label"] |
| 87 | + if y_train.values.ndim == 2 and y_train.values.shape[1] == 1: |
| 88 | + l_name = df_train["label"].columns[0] |
| 89 | + # Convert label into alpha |
| 90 | + df_train["label"][l_name] = df_train["label"][l_name] - df_train["label"][l_name].mean(level=0) |
| 91 | + df_valid["label"][l_name] = df_valid["label"][l_name] - df_valid["label"][l_name].mean(level=0) |
| 92 | + mapping_fn = lambda x: 0 if x < 0 else 1 |
| 93 | + df_train["label_c"] = df_train["label"][l_name].apply(mapping_fn) |
| 94 | + df_valid["label_c"] = df_valid["label"][l_name].apply(mapping_fn) |
| 95 | + x_train, y_train = df_train["feature"], df_train["label_c"].values |
| 96 | + x_valid, y_valid = df_valid["feature"], df_valid["label_c"].values |
| 97 | + else: |
| 98 | + raise ValueError("LightGBM doesn't support multi-label training") |
| 99 | + |
| 100 | + dtrain = lgb.Dataset(x_train.values, label=y_train) |
| 101 | + dvalid = lgb.Dataset(x_valid.values, label=y_valid) |
| 102 | + return dtrain, dvalid |
| 103 | + |
| 104 | + def fit( |
| 105 | + self, |
| 106 | + dataset: DatasetH, |
| 107 | + num_boost_round=1000, |
| 108 | + early_stopping_rounds=50, |
| 109 | + verbose_eval=20, |
| 110 | + evals_result=dict(), |
| 111 | + **kwargs |
| 112 | + ): |
| 113 | + dtrain, dvalid = self._prepare_data(dataset) |
| 114 | + self.model = lgb.train( |
| 115 | + self.params, |
| 116 | + dtrain, |
| 117 | + num_boost_round=num_boost_round, |
| 118 | + valid_sets=[dtrain, dvalid], |
| 119 | + valid_names=["train", "valid"], |
| 120 | + early_stopping_rounds=early_stopping_rounds, |
| 121 | + verbose_eval=verbose_eval, |
| 122 | + evals_result=evals_result, |
| 123 | + **kwargs |
| 124 | + ) |
| 125 | + evals_result["train"] = list(evals_result["train"].values())[0] |
| 126 | + evals_result["valid"] = list(evals_result["valid"].values())[0] |
| 127 | + |
| 128 | + def predict(self, dataset): |
| 129 | + if self.model is None: |
| 130 | + raise ValueError("model is not fitted yet!") |
| 131 | + x_test = dataset.prepare("test", col_set="feature", data_key=DataHandlerLP.DK_I) |
| 132 | + return pd.Series(self.model.predict(x_test.values), index=x_test.index) |
| 133 | + |
| 134 | + def finetune(self, dataset: DatasetH, num_boost_round=10, verbose_eval=20): |
| 135 | + """ |
| 136 | + finetune model |
| 137 | +
|
| 138 | + Parameters |
| 139 | + ---------- |
| 140 | + dataset : DatasetH |
| 141 | + dataset for finetuning |
| 142 | + num_boost_round : int |
| 143 | + number of round to finetune model |
| 144 | + verbose_eval : int |
| 145 | + verbose level |
| 146 | + """ |
| 147 | + # Based on existing model and finetune by train more rounds |
| 148 | + dtrain, _ = self._prepare_data(dataset) |
| 149 | + self.model = lgb.train( |
| 150 | + self.params, |
| 151 | + dtrain, |
| 152 | + num_boost_round=num_boost_round, |
| 153 | + init_model=self.model, |
| 154 | + valid_sets=[dtrain], |
| 155 | + valid_names=["train"], |
| 156 | + verbose_eval=verbose_eval, |
| 157 | + ) |
0 commit comments