{"cells":[{"metadata":{},"cell_type":"markdown","source":"# Riiid Neural Oblivious Decision Ensembles\n\nFrom paper [Neural Oblivious Decision Ensembles for Deep Learning on Tabular Data][1]. \n\nGithub (Pytorch): https://github.com/Qwicen/node\n\nKaggle (Tensorflow): https://www.kaggle.com/marcusgawronsky/differentiable-catboost-node-in-tensorflow-2-0\n\n![image.png](attachment:image.png)\n\n[1]: https://arxiv.org/pdf/1909.06312.pdf","attachments":{"image.png":{"image/png":"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"}}},{"metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","trusted":true},"cell_type":"code","source":"# useful\nimport random, os, math, sys, gc, datetime\nimport numpy as np\nimport pandas as pd\nfrom sklearn.model_selection import StratifiedKFold\nfrom sklearn.metrics import roc_auc_score\nfrom tqdm.notebook import tqdm\nfrom typing import Union, Optional\nfrom time import time\nimport warnings\nwarnings.filterwarnings('ignore')\n\n# neural nets\nimport tensorflow as tf\nimport tensorflow.keras.backend as K\nfrom tensorflow.keras.callbacks import ModelCheckpoint, ReduceLROnPlateau, EarlyStopping\nimport tensorflow_addons as tfa\nimport tensorflow_probability as tfp\n\n# custom\nimport riiideducation","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"### CONSTANTS"},{"metadata":{"trusted":true},"cell_type":"code","source":"# PIVOT DATAFRAMES\npiv1 = pd.read_csv(\"../input/riiid-fixed-infos/content.csv\")\npiv2 = pd.read_csv(\"../input/riiid-fixed-infos/task.csv\")\npiv3 = pd.read_csv(\"../input/riiid-fixed-infos/user.csv\")\n\nfor col, df in zip([\"content_sum\", \"task_container_sum\", \"user_sum\"], [piv1, piv2, piv3]):\n    df[col] = (df[col] - df[col].min()) / (df[col].max() - df[col].min())\n#\nm1 = piv1[\"content_sum\"].median()\nm2 = piv2[\"task_container_sum\"].median()\nm3 = piv3[\"user_sum\"].median()\n\n\n# OTHER CONSTABTS\nTARGET = \"answered_correctly\"\nTIME_MEAN = 21000.0\nTIME_MIN = 0.0\nTIME_MAX = 300000.0\nmap_prior = {True:1, False:0}","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"def preprocess(df):\n    df = df.merge(piv1, how=\"left\", on=\"content_id\")\n    df[\"content_emb\"] = df[\"content_emb\"].fillna(0.5)\n    df[\"content_sum\"] = df[\"content_sum\"].fillna(m1)\n    df = df.merge(piv2, how=\"left\", on=\"task_container_id\")\n    df[\"task_container_emb\"] = df[\"task_container_emb\"].fillna(0.5)\n    df[\"task_container_sum\"] = df[\"task_container_sum\"].fillna(m2)\n    df = df.merge(piv3, how=\"left\", on=\"user_id\")\n    df[\"user_emb\"] = df[\"user_emb\"].fillna(0.5)\n    df[\"user_sum\"] = df[\"user_sum\"].fillna(m3)\n    df[\"prior_question_elapsed_time\"] = df[\"prior_question_elapsed_time\"].fillna(TIME_MEAN)\n    df[\"duration\"] = (df[\"prior_question_elapsed_time\"] - TIME_MIN) / (TIME_MAX - TIME_MIN)\n    df[\"prior_answer\"] = df[\"prior_question_had_explanation\"].map(map_prior)\n    df[\"prior_answer\"] = df[\"prior_answer\"].fillna(0.5)\n    #df = df.fillna(-1)\n    epsilon = 1e-6\n    df[\"score\"] = 2*df[\"content_emb\"]*df[\"user_emb\"] / (df[\"content_emb\"]+ df[\"user_emb\"] + epsilon)\n    return df\n#=========","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"## TRAINING"},{"metadata":{"trusted":true},"cell_type":"code","source":"%%time\ntr = pd.read_csv(\"../input/riiid-test-answer-prediction/train.csv\", \n                 low_memory=False, nrows=10**7)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"%%time\ntr = preprocess(tr)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"FE = [\"content_emb\",\"content_sum\" ,\"task_container_emb\", \"task_container_sum\",\n      \"user_emb\", \"user_sum\",\"duration\", \"prior_answer\",\"score\"]","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"x = tr.loc[tr.answered_correctly!=-1, FE].values\ny = tr.loc[tr.answered_correctly!=-1, TARGET].values","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# ODST"},{"metadata":{"_kg_hide-input":false,"trusted":true},"cell_type":"code","source":"@tf.function\ndef sparsemoid(inputs: tf.Tensor):\n    return tf.clip_by_value(0.5 * inputs + 0.5, 0., 1.)\n\n@tf.function\ndef identity(x: tf.Tensor):\n    return x\n\nclass ODST(tf.keras.layers.Layer):\n    def __init__(self, n_trees: int = 3, depth: int = 4, units: int = 1, threshold_init_beta: float = 1., **kwargs):\n        super(ODST, self).__init__()\n        self.initialized = False\n        self.n_trees = n_trees\n        self.depth = depth\n        self.units = units\n        self.threshold_init_beta = threshold_init_beta\n    \n    def build(self, input_shape: tf.TensorShape):\n        feature_selection_logits_init = tf.zeros_initializer()\n        self.feature_selection_logits = tf.Variable(initial_value=feature_selection_logits_init(shape=(input_shape[-1], self.n_trees, self.depth), dtype='float32'),\n                                 trainable=True)        \n        \n        feature_thresholds_init = tf.zeros_initializer()\n        self.feature_thresholds = tf.Variable(initial_value=feature_thresholds_init(shape=(self.n_trees, self.depth), dtype='float32'),\n                                 trainable=True)\n        \n        log_temperatures_init = tf.ones_initializer()\n        self.log_temperatures = tf.Variable(initial_value=log_temperatures_init(shape=(self.n_trees, self.depth), dtype='float32'),\n                                 trainable=True)\n        \n        indices = tf.keras.backend.arange(0, 2 ** self.depth, 1)\n        offsets = 2 ** tf.keras.backend.arange(0, self.depth, 1)\n        bin_codes = (tf.reshape(indices, (1, -1)) // tf.reshape(offsets, (-1, 1)) % 2)\n        bin_codes_1hot = tf.stack([bin_codes, 1 - bin_codes], axis=-1)\n        self.bin_codes_1hot = tf.Variable(initial_value=tf.cast(bin_codes_1hot, 'float32'), \n                                          trainable=False)\n        \n        response_init = tf.ones_initializer()\n        self.response = tf.Variable(initial_value=response_init(shape=(self.n_trees, self.units, 2**self.depth), dtype='float32'), \n                                    trainable=True)\n                \n    def initialize(self, inputs):        \n        feature_values = self.feature_values(inputs)\n        \n        # intialize feature_thresholds\n        percentiles_q = (100 * tfp.distributions.Beta(self.threshold_init_beta, \n                                                      self.threshold_init_beta)\n                         .sample([self.n_trees * self.depth]))\n        flattened_feature_values = tf.map_fn(tf.keras.backend.flatten, feature_values)\n        init_feature_thresholds = tf.linalg.diag_part(tfp.stats.percentile(flattened_feature_values, percentiles_q, axis=0))\n        \n        self.feature_thresholds.assign(tf.reshape(init_feature_thresholds, self.feature_thresholds.shape))\n        \n        \n        # intialize log_temperatures\n        self.log_temperatures.assign(tfp.stats.percentile(tf.math.abs(feature_values - self.feature_thresholds), 50, axis=0))\n        \n        \n        \n    def feature_values(self, inputs: tf.Tensor, training: bool = None):\n        feature_selectors = tfa.activations.sparsemax(self.feature_selection_logits)\n        # ^--[in_features, n_trees, depth]\n\n        feature_values = tf.einsum('bi,ind->bnd', inputs, feature_selectors)\n        # ^--[batch_size, n_trees, depth]\n        \n        return feature_values\n        \n    def call(self, inputs: tf.Tensor, training: bool = None):\n        if not self.initialized:\n            self.initialize(inputs)\n            self.initialized = True\n            \n        feature_values = self.feature_values(inputs)\n        \n        threshold_logits = (feature_values - self.feature_thresholds) * tf.math.exp(-self.log_temperatures)\n\n        threshold_logits = tf.stack([-threshold_logits, threshold_logits], axis=-1)\n        # ^--[batch_size, n_trees, depth, 2]\n\n        bins = sparsemoid(threshold_logits)\n        # ^--[batch_size, n_trees, depth, 2], approximately binary\n\n        bin_matches = tf.einsum('btds,dcs->btdc', bins, self.bin_codes_1hot)\n        # ^--[batch_size, n_trees, depth, 2 ** depth]\n\n        response_weights = tf.math.reduce_prod(bin_matches, axis=-2)\n        # ^-- [batch_size, n_trees, 2 ** depth]\n\n        response = tf.einsum('bnd,ncd->bnc', response_weights, self.response)\n        # ^-- [batch_size, n_trees, units]\n        \n        return tf.reduce_sum(response, axis=1)\n    \nclass NODE(tf.keras.Model):\n    def __init__(self, units: int = 1, n_layers: int = 1, dropout_rate = 0.1, \n                 link: tf.function = tf.identity, n_trees: int = 3, depth: int = 4, \n                 threshold_init_beta: float = 1., feature_column: Optional[tf.keras.layers.DenseFeatures] = None, \n                 **kwargs):\n        super(NODE, self).__init__()\n        self.units = units\n        self.n_layers = n_layers\n        self.n_trees = n_trees\n        self.depth = depth\n        self.units = units\n        self.threshold_init_beta = threshold_init_beta\n        self.feature_column = feature_column\n        self.dropout_rate = dropout_rate\n        \n        if feature_column is None:\n            self.feature = tf.keras.layers.Lambda(identity)\n        else:\n            self.feature = feature_column\n        \n        self.bn = tf.keras.layers.BatchNormalization()\n        self.dropout = tf.keras.layers.Dropout(self.dropout_rate)\n        self.ensemble = [ODST(n_trees = n_trees,\n                              depth = depth,\n                              units = units,\n                              threshold_init_beta = threshold_init_beta) \n                         for _ in range(n_layers)]\n        \n        self.link = link\n        \n    def call(self, inputs, training=None):\n        X = self.feature(inputs)\n        X = self.bn(X, training=training)\n        X = self.dropout(X, training=training)\n        \n        for i, tree in enumerate(self.ensemble):\n            H = tree(X)\n            X = tf.concat([X, H], axis=1)\n            \n        return self.link(H)","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# Model"},{"metadata":{"trusted":true},"cell_type":"code","source":"def create_NODE(n_layers, units, dropout_rate, depth, n_trees, link, learning_rate):\n    \n    node = NODE(n_layers = n_layers, units = units, dropout_rate = dropout_rate, \n                depth = depth, n_trees = n_trees, link = link)\n    \n    node.compile(optimizer = tf.keras.optimizers.Adam(learning_rate = learning_rate), \n                 metrics = [tf.keras.metrics.AUC(name = 'auc')], \n                 loss = 'binary_crossentropy')\n    \n    return node","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# TRAIN"},{"metadata":{"trusted":true},"cell_type":"code","source":"N_STARTS = 1\nN_SPLITS = 10\nEPOCHS = 100\nBATCH_SIZE = 30000\nVERBOSE = 0\n\ntrain = x\ntrain_targets = y\n\nres = np.zeros(train_targets.shape[0])\n\nfor seed in range(N_STARTS):\n    start_time_seed = time()\n    K.clear_session()\n    tf.random.set_seed(seed)\n    mean_score = 0\n    skf = StratifiedKFold(n_splits = N_SPLITS, random_state = seed, shuffle = True)\n    for n, (tr_idx, te_idx) in enumerate(skf.split(train_targets, train_targets)):\n        \n        start_time_fold = time()\n        x_tr, x_val = train[tr_idx], train[te_idx]\n        y_tr, y_val = train_targets[tr_idx], train_targets[te_idx]\n            \n        model = create_NODE(n_layers = 1, units = 1, dropout_rate = 0.3, depth = 6, \n                            n_trees = 16, link = tf.keras.activations.sigmoid, learning_rate = 1e-3)\n        ckp = ModelCheckpoint(f'NODE_{seed}_{n}', monitor = 'val_auc', verbose = VERBOSE, \n                              save_best_only = True, save_weights_only = True, mode = 'max')\n        rlr = ReduceLROnPlateau(monitor = 'val_auc', factor = 0.1, patience = 3, \n                                verbose = VERBOSE, min_delta = 1e-4, mode = 'max')\n        es = EarlyStopping(monitor = 'val_auc', min_delta = 1e-4, patience = 5, mode = 'max', \n                           baseline = None, restore_best_weights = True, verbose = VERBOSE)\n        history = model.fit(x_tr, y_tr, validation_data = (x_val, y_val), epochs = EPOCHS, \n                            batch_size = BATCH_SIZE, callbacks = [ckp, rlr, es], verbose = VERBOSE)\n        hist = pd.DataFrame(history.history)\n        fold_score = hist['val_auc'].max()\n        mean_score += fold_score / N_SPLITS\n        model.load_weights(f'NODE_{seed}_{n}')\n        val_predict = model.predict(x_val, batch_size = BATCH_SIZE * 4)[:, 0]\n        \n        res[te_idx] += val_predict / N_STARTS\n        print(f'[{str(datetime.timedelta(seconds = time() - start_time_fold))[2:7]}] NODE Seed {seed}, Fold {n}:', fold_score)\n        break\n    break\n        \n#     print(f'[{str(datetime.timedelta(seconds = time() - start_time_seed))[2:7]}] NODE Seed {seed} Mean Score:', mean_score)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"# print(f'NODE OOF Metric: {roc_auc_score(train_targets, res)}')","execution_count":null,"outputs":[]},{"metadata":{},"cell_type":"markdown","source":"# PREDICTION"},{"metadata":{"trusted":true},"cell_type":"code","source":"env = riiideducation.make_env()\niter_test = env.iter_test()","execution_count":null,"outputs":[]},{"metadata":{"trusted":true},"cell_type":"code","source":"model = create_NODE(n_layers = 1, units = 1, dropout_rate = 0.3, depth = 6, \n                    n_trees = 16, link = tf.keras.activations.sigmoid, learning_rate = 1e-3)\n\n# it = 0\nfor test_df, sample_prediction_df in iter_test:\n#     it += 1\n#     if it % 100 == 0:\n#         print(it)\n    test_df = preprocess(test_df)\n    x_te = test_df[FE].values\n    model.load_weights(f'NODE_0_0')\n    test_df['answered_correctly'] = model.predict(x_te, batch_size = 50_000, verbose = 0)[:, 0]\n    env.predict(test_df.loc[test_df['content_type_id'] == 0, ['row_id', 'answered_correctly']])\n#=================================================\n# print(it)","execution_count":null,"outputs":[]},{"metadata":{"trusted":true,"_kg_hide-output":true},"cell_type":"code","source":"# it = 0\n# for test_df, sample_prediction_df in iter_test:\n#     it += 1\n#     if it % 100 == 0:\n#         print(it)\n#     test_df = preprocess(test_df)\n#     x_te = test_df[FE].values\n#     test_df['answered_correctly'] = 0\n#     for seed in range(N_STARTS):\n#         for fold in range(N_SPLITS):\n#             model = create_NODE(n_layers = 1, units = 1, dropout_rate = 0.3, depth = 6, \n#                                 n_trees = 16, link = tf.keras.activations.sigmoid, learning_rate = 1e-3)\n#             model.load_weights(f'NODE_{seed}_{fold}')\n#             test_df['answered_correctly'] += model.predict(x_te, batch_size = 50_000, verbose = 0)[:, 0] / (N_STARTS * N_SPLITS)\n#     env.predict(test_df.loc[test_df['content_type_id'] == 0, ['row_id', 'answered_correctly']])\n# #=================================================\n# print(it)","execution_count":null,"outputs":[]}],"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"pygments_lexer":"ipython3","nbconvert_exporter":"python","version":"3.6.4","file_extension":".py","codemirror_mode":{"name":"ipython","version":3},"name":"python","mimetype":"text/x-python"}},"nbformat":4,"nbformat_minor":4}