849 lines
35 KiB
Plaintext
849 lines
35 KiB
Plaintext
{
|
|
"cells": [
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import pandas as pd\n",
|
|
"from IPython.display import clear_output, Markdown, Math\n",
|
|
"import ipywidgets as widgets\n",
|
|
"import os\n",
|
|
"import glob\n",
|
|
"import json\n",
|
|
"import numpy as np\n",
|
|
"import itertools\n",
|
|
"import sklearn.utils as sku\n",
|
|
"from sklearn.feature_extraction.text import CountVectorizer, TfidfVectorizer, HashingVectorizer\n",
|
|
"from sklearn.model_selection import train_test_split\n",
|
|
"from sklearn.preprocessing import MultiLabelBinarizer, LabelBinarizer\n",
|
|
"from sklearn.cluster import KMeans\n",
|
|
"import nltk\n",
|
|
"from keras.models import load_model\n",
|
|
"from sklearn.externals import joblib\n",
|
|
"import pickle\n",
|
|
"import operator\n",
|
|
"from sklearn.pipeline import Pipeline\n",
|
|
"import json\n",
|
|
"nltk.download('punkt')\n",
|
|
"nltk.download('averaged_perceptron_tagger')\n",
|
|
"nltk.download('wordnet')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import sys\n",
|
|
"sys.path.append(\"..\")\n",
|
|
"\n",
|
|
"import Tools.Emoji_Distance as edist\n",
|
|
"\n",
|
|
"def emoji2sent(emoji_arr, only_emoticons=True):\n",
|
|
" return np.array([edist.emoji_to_sentiment_vector(e, only_emoticons=only_emoticons) for e in emoji_arr])\n",
|
|
"\n",
|
|
"def sent2emoji(sent_arr, custom_target_emojis=None, only_emoticons=True):\n",
|
|
" return [edist.sentiment_vector_to_emoji(s, custom_target_emojis=custom_target_emojis, only_emoticons=only_emoticons) for s in sent_arr]"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"SINGLE_LABEL = True"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"----\n",
|
|
"## classes and functions we are using later:\n",
|
|
"----"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* functions for selecting items from a set / list"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"def latest(lst):\n",
|
|
" return lst[-1] if len(lst) > 0 else 'X' \n",
|
|
"def most_common(lst):\n",
|
|
" # trying to find the most common used emoji in the given lst\n",
|
|
" return max(set(lst), key=lst.count) if len(lst) > 0 else \"X\" # setting label to 'X' if there is an empty emoji list"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* our emoji blacklist (skin and sex modifiers)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"# defining blacklist for modifier emojis:\n",
|
|
"emoji_blacklist = set([\n",
|
|
" chr(0x1F3FB),\n",
|
|
" chr(0x1F3FC),\n",
|
|
" chr(0x1F3FD),\n",
|
|
" chr(0x1F3FE),\n",
|
|
" chr(0x1F3FF),\n",
|
|
" chr(0x2642),\n",
|
|
" chr(0x2640)\n",
|
|
"])"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* lemmatization helper functions"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"from nltk.stem.snowball import SnowballStemmer\n",
|
|
"from nltk.stem import WordNetLemmatizer\n",
|
|
"from nltk import pos_tag\n",
|
|
"from nltk import word_tokenize\n",
|
|
"from nltk.corpus import wordnet\n",
|
|
"\n",
|
|
"def get_wordnet_pos(treebank_tag):\n",
|
|
"\n",
|
|
" if treebank_tag.startswith('J'):\n",
|
|
" return wordnet.ADJ\n",
|
|
" elif treebank_tag.startswith('V'):\n",
|
|
" return wordnet.VERB\n",
|
|
" elif treebank_tag.startswith('N'):\n",
|
|
" return wordnet.NOUN\n",
|
|
" elif treebank_tag.startswith('R'):\n",
|
|
" return wordnet.ADV\n",
|
|
" else:\n",
|
|
" return wordnet.NOUN"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"### sample data manager\n",
|
|
"the sample data manager loads and preprocesses data\n",
|
|
"most common way to use:\n",
|
|
"\n",
|
|
"\n",
|
|
"* `sdm = sample_data_manager.generate_and_read(path:str, only_emoticons=True, apply_stemming=True, n_top_emojis=-1, file_range=None)`\n",
|
|
"\n",
|
|
" * Generates a sample_data_manager object and preprocess data in one step\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class sample_data_manager(object):\n",
|
|
" @staticmethod\n",
|
|
" def generate_and_read(path:str, only_emoticons=True, apply_stemming=True, n_top_emojis=-1, file_range=None, n_kmeans_cluster=-1, progress_callback=None):\n",
|
|
" \"\"\"\n",
|
|
" generate, read and process train data in one step.\n",
|
|
" \n",
|
|
" @param path: folder containing json files to process\n",
|
|
" @param only_emoticons: if True, only messages containing emoticons (provided by Tools.Emoji_Distance) are used\n",
|
|
" @param apply_stemming: apply stemming and lemmatization on dataset\n",
|
|
" @param n_top_emojis: only use messages containing one of <`n_top_emojis`>-top emojis. set to `-1` to prevent top emoji filtering\n",
|
|
" @param file_range: range of file's indices to read (eg `range(3)` to read the first three files). If `None`: all files are read\n",
|
|
" @param n_kmeans_cluster: generating multilabeled labels with kmeans with these number of clusters. Set to -1 to use the plain sentiment space as label\n",
|
|
" \n",
|
|
" @return: sample_data_manager object\n",
|
|
" \"\"\"\n",
|
|
" sdm = sample_data_manager(path)\n",
|
|
" sdm.read_files(file_index_range=range(sdm.n_files) if file_range is None else file_range, only_emoticons=only_emoticons, progress_callback=progress_callback)\n",
|
|
" if apply_stemming:\n",
|
|
" sdm.apply_stemming_and_lemmatization()\n",
|
|
" \n",
|
|
" sdm.generate_emoji_count_and_weights()\n",
|
|
" \n",
|
|
" if n_top_emojis > 0:\n",
|
|
" sdm.filter_by_top_emojis(n_top=n_top_emojis)\n",
|
|
" \n",
|
|
" if n_kmeans_cluster > 0:\n",
|
|
" sdm.generate_kmeans_binary_label(only_emoticons=only_emoticons, n_clusters=n_kmeans_cluster)\n",
|
|
" \n",
|
|
" return sdm\n",
|
|
" \n",
|
|
" \n",
|
|
" def __init__(self, data_root_folder:str):\n",
|
|
" \"\"\"\n",
|
|
" constructor for manual initialization\n",
|
|
" \n",
|
|
" @param data_root_folder: folder containing json files to process\n",
|
|
" \"\"\"\n",
|
|
" self.data_root_folder = data_root_folder\n",
|
|
" self.json_files = sorted(glob.glob(self.data_root_folder + \"/*.json\"))\n",
|
|
" self.n_files = len(self.json_files)\n",
|
|
" self.raw_data = None\n",
|
|
" self.emojis = None\n",
|
|
" self.plain_text = None\n",
|
|
" self.labels = None\n",
|
|
" self.emoji_count = None\n",
|
|
" self.emoji_weights = None\n",
|
|
" self.X = None\n",
|
|
" self.y = None\n",
|
|
" self.Xt = None\n",
|
|
" self.yt = None\n",
|
|
" self.top_emojis = None\n",
|
|
" self.binary_labels = None\n",
|
|
" self.use_binary_labels = False\n",
|
|
" self.kmeans_cluster = None\n",
|
|
" self.label_binarizer = None\n",
|
|
" \n",
|
|
" def read_files(self, file_index_range:list, only_emoticons=True, progress_callback=None):\n",
|
|
" \"\"\"\n",
|
|
" reading (multiple) files to one panda table.\n",
|
|
" \n",
|
|
" @param file_index_range: range of file's indices to read (eg `range(3)` to read the first three files)\n",
|
|
" @param only_emoticons: if True, only messages containing emoticons (aka smileys) are used. This classification is derived from Tools.Emoji_Distance\n",
|
|
" \"\"\"\n",
|
|
" assert np.min(file_index_range) >= 0 and np.max(file_index_range) < self.n_files\n",
|
|
" for i in file_index_range:\n",
|
|
" print(\"reading file: \" + self.json_files[i] + \"...\")\n",
|
|
" if self.raw_data is None:\n",
|
|
" self.raw_data = pd.read_json(self.json_files[i], encoding=\"utf-8\")\n",
|
|
" else:\n",
|
|
" self.raw_data = self.raw_data.append(pd.read_json(self.json_files[i], encoding=\"utf-8\"))\n",
|
|
" if progress_callback is not None:\n",
|
|
" progress_callback()\n",
|
|
" self.emojis = self.raw_data['EMOJI']\n",
|
|
" self.plain_text = self.raw_data['text']\n",
|
|
" \n",
|
|
" # replacing keywords. TODO: maybe these information can be extracted and used\n",
|
|
" self.plain_text = self.plain_text.str.replace(\"(<EMOJI>|<USER>|<HASHTAG>)\",\"\").str.replace(\"[\" + \"\".join(list(emoji_blacklist)) + \"]\",\"\")\n",
|
|
" \n",
|
|
" # so far filtering for the latest emoji. TODO: maybe there are also better approaches\n",
|
|
" self.labels = emoji2sent([latest(e) for e in self.emojis], only_emoticons=only_emoticons )\n",
|
|
" \n",
|
|
" # and filter out all samples we have no label for:\n",
|
|
" wrong_labels = np.isnan(np.linalg.norm(self.labels, axis=1)) \n",
|
|
"\n",
|
|
" self.labels = self.labels[np.invert(wrong_labels)]\n",
|
|
" self.plain_text = self.plain_text[np.invert(wrong_labels)]\n",
|
|
" self.emojis = self.emojis[np.invert(wrong_labels)]\n",
|
|
" \n",
|
|
" print(\"imported \" + str(len(self.labels)) + \" samples\")\n",
|
|
" \n",
|
|
" def apply_stemming_and_lemmatization(self):\n",
|
|
" \"\"\"\n",
|
|
" apply stemming and lemmatization to plain text samples\n",
|
|
" \"\"\"\n",
|
|
" stemmer = SnowballStemmer(\"english\")\n",
|
|
" for key in self.plain_text.keys():\n",
|
|
" stemmed_sent = []\n",
|
|
" for word in self.plain_text[key].split(\" \"):\n",
|
|
" word_stemmed = stemmer.stem(word)\n",
|
|
" stemmed_sent.append(word_stemmed)\n",
|
|
" stemmed_sent = (\" \").join(stemmed_sent)\n",
|
|
" self.plain_text[key] = stemmed_sent\n",
|
|
" \n",
|
|
" lemmatizer = WordNetLemmatizer()\n",
|
|
" for key in self.plain_text.keys():\n",
|
|
" lemmatized_sent = []\n",
|
|
" sent_pos = pos_tag(word_tokenize(self.plain_text[key]))\n",
|
|
" for word in sent_pos:\n",
|
|
" wordnet_pos = get_wordnet_pos(word[1].lower())\n",
|
|
" word_lemmatized = lemmatizer.lemmatize(word[0], pos=wordnet_pos)\n",
|
|
" lemmatized_sent.append(word_lemmatized)\n",
|
|
" lemmatized_sent = (\" \").join(lemmatized_sent)\n",
|
|
" self.plain_text[key] = lemmatized_sent\n",
|
|
" \n",
|
|
" def generate_emoji_count_and_weights(self):\n",
|
|
" \"\"\"\n",
|
|
" counting occurences of emojis\n",
|
|
" \"\"\"\n",
|
|
" self.emoji_count = {}\n",
|
|
" for e_list in self.emojis:\n",
|
|
" for e in set(e_list):\n",
|
|
" if e not in self.emoji_count:\n",
|
|
" self.emoji_count[e] = 0\n",
|
|
" self.emoji_count[e] += 1\n",
|
|
" \n",
|
|
" emoji_sum = sum([self.emoji_count[e] for e in self.emoji_count])\n",
|
|
"\n",
|
|
" self.emoji_weights = {}\n",
|
|
" for e in self.emoji_count:\n",
|
|
" # tfidf for emojis\n",
|
|
" self.emoji_weights[e] = np.log((emoji_sum / self.emoji_count[e]))\n",
|
|
"\n",
|
|
" weights_sum= sum([self.emoji_weights[x] for x in self.emoji_weights])\n",
|
|
"\n",
|
|
" # normalize:\n",
|
|
" for e in self.emoji_weights:\n",
|
|
" self.emoji_weights[e] = self.emoji_weights[e] / weights_sum\n",
|
|
"\n",
|
|
" self.emoji_weights['X'] = 0 # dummy values\n",
|
|
" self.emoji_count['X'] = 0\n",
|
|
" \n",
|
|
" def get_emoji_count(self):\n",
|
|
" \"\"\"\n",
|
|
" @return: descending list of tuples in form (<emoji as character>, <emoji count>) \n",
|
|
" \"\"\"\n",
|
|
" assert self.emoji_count is not None\n",
|
|
" \n",
|
|
" sorted_emoji_count = list(reversed(sorted(self.emoji_count.items(), key=operator.itemgetter(1))))\n",
|
|
" #display(sorted_emoji_count)\n",
|
|
" return sorted_emoji_count\n",
|
|
" \n",
|
|
" def filter_by_top_emojis(self,n_top = 20):\n",
|
|
" \"\"\"\n",
|
|
" filgter out messages not containing one of the `n_top` emojis\n",
|
|
" \n",
|
|
" @param n_top: number of top emojis used for filtering\n",
|
|
" \"\"\"\n",
|
|
" assert self.labels is not None # ← messages are already read in\n",
|
|
" \n",
|
|
" self.top_emojis = [x[0] for x in self.get_emoji_count()[:n_top]]\n",
|
|
" in_top = [edist.sentiment_vector_to_emoji(x) in self.top_emojis for x in self.labels]\n",
|
|
" self.labels = self.labels[in_top]\n",
|
|
" self.plain_text = self.plain_text[in_top]\n",
|
|
" self.emojis = self.emojis[in_top]\n",
|
|
" print(\"remaining samples after top emoji filtering: \", len(self.labels))\n",
|
|
" \n",
|
|
" def generate_kmeans_binary_label(self, only_emoticons=True, n_clusters=5):\n",
|
|
" \"\"\"\n",
|
|
" generate binary labels using kmeans.\n",
|
|
" \n",
|
|
" @param only_emoticons: set whether we're using the full emoji set or only emoticons\n",
|
|
" @param n_clusters: number of cluster we're generating in emoji's sentiment space\n",
|
|
" \"\"\"\n",
|
|
" assert self.labels is not None\n",
|
|
" array_sentiment_vectors = edist.list_sentiment_emoticon_vectors if only_emoticons else edist.list_sentiment_vectors\n",
|
|
" array_sentiment_vectors = np.array(array_sentiment_vectors)\n",
|
|
" \n",
|
|
" list_emojis = edist.list_emoticon_emojis if only_emoticons else edist.list_emojis\n",
|
|
" self.use_binary_labels = True\n",
|
|
" print(\"clustering following emojis: \" + \"\".join(list_emojis) + \"...\")\n",
|
|
" self.kmeans_cluster = KMeans(n_clusters=n_clusters).fit(array_sentiment_vectors)\n",
|
|
" print(\"clustering done\")\n",
|
|
" self.label_binarizer = LabelBinarizer()\n",
|
|
" \n",
|
|
" multiclass_labels = self.kmeans_cluster.predict(self.labels)\n",
|
|
" \n",
|
|
" # FIXME: we have to guarantee that in every dataset all classes occur.\n",
|
|
" # otherwise batch fitting is not possible!\n",
|
|
" # (or we have to precompute the mlb fitting process somewhere...)\n",
|
|
" self.binary_labels = self.label_binarizer.fit_transform(multiclass_labels)\n",
|
|
" \n",
|
|
" \n",
|
|
" def create_train_test_split(self, split = 0.1, random_state = 4222):\n",
|
|
" assert self.plain_text is not None and self.labels is not None\n",
|
|
" if self.X is not None:\n",
|
|
" sys.stderr.write(\"WARNING: overwriting existing train/test split \\n\")\n",
|
|
" \n",
|
|
" labels = self.binary_labels if self.use_binary_labels else self.labels\n",
|
|
" assert labels is not None\n",
|
|
" self.X, self.Xt, self.y, self.yt = train_test_split(self.plain_text, labels, test_size=split, random_state=random_state)\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* the pipeline manager saves and stores sklearn pipelines. Keras models are handled differently, so the have to be named explicitly during save and load operations"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class pipeline_manager(object):\n",
|
|
" @staticmethod\n",
|
|
" def load_from_pipeline_file(pipeline_file:str):\n",
|
|
" \"\"\"\n",
|
|
" loading a json configuration file and using it's paramters to call 'load_pipeline_from_files'\n",
|
|
" \"\"\"\n",
|
|
" with open(pipeline_file, 'r') as f:\n",
|
|
" d = json.load(f)\n",
|
|
" \n",
|
|
" keras_models = d['keras_models']\n",
|
|
" all_models = d['all_models']\n",
|
|
" \n",
|
|
" return pipeline_manager.load_pipeline_from_files(pipeline_file.rsplit('.',1)[0], keras_models, all_models)\n",
|
|
"\n",
|
|
"\n",
|
|
" @staticmethod\n",
|
|
" def load_pipeline_from_files(file_prefix:str, keras_models = [], all_models = []):\n",
|
|
" \"\"\"\n",
|
|
" load a pipeline from files. A pipeline should be represented by multiple model files in the form '<file_prefix>.<model_name>'\n",
|
|
" \n",
|
|
" @param file_prefix: basename of all files (without extension)\n",
|
|
" @param keras_models: list of keras models (keras model files, only extension name). Leave this list empty if this is not a keras pipeline\n",
|
|
" @param all_models: list of all models (including keras_models, only extension name).\n",
|
|
" \n",
|
|
" @return a pipeline manager object\n",
|
|
" \"\"\"\n",
|
|
" \n",
|
|
" pm = pipeline_manager(keras_models=keras_models)\n",
|
|
" pm.load(file_prefix, all_models)\n",
|
|
" return pm\n",
|
|
" \n",
|
|
" @staticmethod\n",
|
|
" def create_keras_pipeline_with_vectorizer(vectorizer, layers, sdm:sample_data_manager, loss=None, optimizer=None):\n",
|
|
" '''\n",
|
|
" creates pipeline with vectorizer and keras classifier\n",
|
|
" \n",
|
|
" @param vectorizer: Vectorizer object. will be fitted with data provided by sdm\n",
|
|
" @param layers: list of keras layers. One keras layer is a tuple in form: (<#neurons:int>, <activation_func:str>)\n",
|
|
" @param sdm: sample data manager to get data for the vectorizer\n",
|
|
" @param loss: set keras loss function. Depending whether sdm use multiclass labels `categorical_crossentropy` or `mean_squared_error` is used as default\n",
|
|
" @param optimizer: set keras optimizer. Depending whether sdm use multiclass labels `sgd` or `adam` is used as default\n",
|
|
" \n",
|
|
" @return: a pipeline manager object\n",
|
|
" \n",
|
|
" '''\n",
|
|
" from keras.models import Sequential\n",
|
|
" from keras.layers import Dense\n",
|
|
" \n",
|
|
" if sdm.X is None:\n",
|
|
" sdm.create_train_test_split()\n",
|
|
" \n",
|
|
" vec_train = vectorizer.fit_transform(sdm.X)\n",
|
|
" vec_test = vectorizer.transform(sdm.Xt)\n",
|
|
" # creating keras model:\n",
|
|
" model=Sequential()\n",
|
|
" \n",
|
|
" keras_layers = []\n",
|
|
" first_layer = True\n",
|
|
" for layer in layers:\n",
|
|
" if first_layer:\n",
|
|
" model.add(Dense(units=layer[0], activation=layer[1], input_dim=vectorizer.transform([\" \"])[0]._shape[1]))\n",
|
|
" first_layer = False\n",
|
|
" else:\n",
|
|
" model.add(Dense(units=layer[0], activation=layer[1]))\n",
|
|
" \n",
|
|
" if sdm.use_binary_labels: \n",
|
|
" loss_function = loss if loss is not None else 'categorical_crossentropy'\n",
|
|
" optimizer_function = optimizer if optimizer is not None else 'sgd'\n",
|
|
" model.compile(loss=loss_function,\n",
|
|
" optimizer=optimizer_function,\n",
|
|
" metrics=['accuracy'])\n",
|
|
" else:\n",
|
|
" loss_function = loss if loss is not None else 'mean_squared_error'\n",
|
|
" optimizer_function = optimizer if optimizer is not None else 'adam'\n",
|
|
" model.compile(loss=loss_function,\n",
|
|
" optimizer=optimizer_function)\n",
|
|
" \n",
|
|
" pipeline = Pipeline([\n",
|
|
" ('vectorizer',vectorizer),\n",
|
|
" ('keras_model', model)\n",
|
|
" ])\n",
|
|
" \n",
|
|
" return pipeline_manager(pipeline=pipeline, keras_models=['keras_model'])\n",
|
|
" \n",
|
|
" @staticmethod\n",
|
|
" def create_pipeline_with_classifier_and_vectorizer(vectorizer, classifier, sdm:sample_data_manager = None):\n",
|
|
" '''\n",
|
|
" creates pipeline with vectorizer and non-keras classifier\n",
|
|
" \n",
|
|
" @param vectorizer: Vectorizer object. will be fitted with data provided by sdm\n",
|
|
" @param classifier: unfitted classifier object (should be compatible with all sklearn classifiers)\n",
|
|
" @param sdm: sample data manager to get data for the vectorizer\n",
|
|
" \n",
|
|
" @return: a pipeline manager object\n",
|
|
" '''\n",
|
|
" if sdm is not None:\n",
|
|
" if sdm.X is None:\n",
|
|
" sdm.create_train_test_split()\n",
|
|
"\n",
|
|
" vec_train = vectorizer.fit_transform(sdm.X)\n",
|
|
" vec_test = vectorizer.transform(sdm.Xt)\n",
|
|
" \n",
|
|
" pipeline = Pipeline([\n",
|
|
" ('vectorizer',vectorizer),\n",
|
|
" ('classifier', classifier)\n",
|
|
" ])\n",
|
|
" \n",
|
|
" return pipeline_manager(pipeline=pipeline, keras_models=[])\n",
|
|
" \n",
|
|
" def __init__(self, pipeline = None, keras_models = []):\n",
|
|
" \"\"\"\n",
|
|
" constructor\n",
|
|
" \n",
|
|
" @param pipeline: a sklearn pipeline\n",
|
|
" @param keras_models: list of keras steps in pipeline. Neccessary because saving and loading from keras models differs from the scikit ones\n",
|
|
" \"\"\"\n",
|
|
" \n",
|
|
" self.pipeline = pipeline\n",
|
|
" self.additional_objects = {}\n",
|
|
" self.keras_models = keras_models\n",
|
|
" \n",
|
|
" def save(self, prefix:str):\n",
|
|
" \"\"\"\n",
|
|
" saving the pipeline. It generates one file per model in the form: '<prefix>.<model_name>'\n",
|
|
" \n",
|
|
" @param prefix: file prefix for all models\n",
|
|
" \"\"\"\n",
|
|
" \n",
|
|
"\n",
|
|
" print(self.keras_models)\n",
|
|
" # doing this like explained here: https://stackoverflow.com/a/43415459\n",
|
|
" for step in self.pipeline.named_steps:\n",
|
|
" if step in self.keras_models:\n",
|
|
" self.pipeline.named_steps[step].model.save(prefix + \".\" + step)\n",
|
|
" else:\n",
|
|
" joblib.dump(self.pipeline.named_steps[step], prefix + \".\" + str(step))\n",
|
|
" \n",
|
|
" load_command = \"pipeline_manager.load_pipeline_from_files( '\"\n",
|
|
" load_command += prefix + \"', \" + str(self.keras_models) + \", \"\n",
|
|
" load_command += str(list(self.pipeline.named_steps.keys())) + \")\"\n",
|
|
"\n",
|
|
" with open(prefix + '.pipeline', 'w') as outfile:\n",
|
|
" json.dump({'keras_models': self.keras_models, 'all_models': [step for step in self.pipeline.named_steps]}, outfile)\n",
|
|
" \n",
|
|
" import __main__ as main\n",
|
|
" if not hasattr(main, '__file__'):\n",
|
|
" display(\"saved pipeline. It can be loaded the following way:\")\n",
|
|
" display(Markdown(\"> ```\\n\"+load_command+\"\\n```\")) # ← if we're in jupyter, print the fancy way :)\n",
|
|
" else:\n",
|
|
" print(\"saved pipeline. It can be loaded the following way:\")\n",
|
|
" print(load_command)\n",
|
|
" \n",
|
|
" \n",
|
|
" def load(self, prefix:str, models = []):\n",
|
|
" \"\"\"\n",
|
|
" load a pipeline. A pipeline should be represented by multiple model files in the form '<prefix>.<model_name>'\n",
|
|
" NOTE: keras model names (if there are some) have to be defined in self.keras_models first!\n",
|
|
" \n",
|
|
" @param prefix: the prefix for all model files\n",
|
|
" @param models: model_names to load\n",
|
|
" \"\"\"\n",
|
|
" self.pipeline = None\n",
|
|
" model_list = []\n",
|
|
" for model in models:\n",
|
|
" if model in self.keras_models:\n",
|
|
" model_list.append((model, load_model(prefix + \".\" + model)))\n",
|
|
" else:\n",
|
|
" model_list.append((model, joblib.load(prefix+\".\" + model)))\n",
|
|
" self.pipeline = Pipeline(model_list)\n",
|
|
" \n",
|
|
" def fit(self,X,y):\n",
|
|
" \"\"\"fitting the pipeline\"\"\"\n",
|
|
" self.pipeline.fit(X,y)\n",
|
|
" \n",
|
|
" def predict(self,X):\n",
|
|
" \"\"\"predict\"\"\"\n",
|
|
" return self.pipeline.predict(X)\n",
|
|
" "
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* the trainer class passes Data from the sample manager to the pipeline manager"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"class trainer(object):\n",
|
|
" def __init__(self, sdm:sample_data_manager, pm:pipeline_manager):\n",
|
|
" \"\"\"constructor\"\"\"\n",
|
|
" self.sdm = sdm\n",
|
|
" self.pm = pm\n",
|
|
" \n",
|
|
" def fit(self, max_size=10000, disabled_fit_steps=['vectorizer'], keras_batch_fitting_layer=['keras_model'], batch_size=None, n_epochs=1, progress_callback=None):\n",
|
|
" \"\"\"\n",
|
|
" fitting data in the pipeline. Because we don't want to refit the vectorizer, the pipeline models containing the vectorizer have to be named explicitly\n",
|
|
" \n",
|
|
" @param max_size: don't train more examples than that number\n",
|
|
" @param disabled_fit_steps: list of pipeline steps that we want to prevent to refit. Normally all vectorizer steps\n",
|
|
" \"\"\"\n",
|
|
" # TODO: make batch fitting available here (eg: continous waiting for data and fitting them)\n",
|
|
" if self.sdm.X is None:\n",
|
|
" self.sdm.create_train_test_split()\n",
|
|
" disabled_fits = {}\n",
|
|
" disabled_fit_transforms = {}\n",
|
|
" \n",
|
|
" disabled_keras_fits = {}\n",
|
|
" \n",
|
|
" named_steps = self.pm.pipeline.named_steps\n",
|
|
" \n",
|
|
" for s in disabled_fit_steps:\n",
|
|
" # now it gets really dirty:\n",
|
|
" # replace fit functions we don't want to call again (e.g. for vectorizers)\n",
|
|
" disabled_fits[s] = named_steps[s].fit\n",
|
|
" disabled_fit_transforms[s] = named_steps[s].fit_transform\n",
|
|
" named_steps[s].fit = lambda self, X, y=None: self\n",
|
|
" named_steps[s].fit_transform = named_steps[s].transform\n",
|
|
" \n",
|
|
" for k in keras_batch_fitting_layer:\n",
|
|
" # forcing batch fitting on keras\n",
|
|
" disabled_keras_fits[k]=named_steps[k].fit\n",
|
|
" named_steps[k].fit = lambda X, y: named_steps[k].train_on_batch(X.todense(), y) # ← why has keras no sparse support on batch progressing!?!?!\n",
|
|
" \n",
|
|
" if batch_size is None:\n",
|
|
" self.pm.fit(X = self.sdm.X[:max_size], y = self.sdm.y[:max_size])\n",
|
|
" else:\n",
|
|
" n = len(self.sdm.X) // batch_size\n",
|
|
" for i in range(n_epochs):\n",
|
|
" for j in range(n):\n",
|
|
" self.pm.fit(X = np.array(self.sdm.X[j*batch_size:(j+1)*batch_size]), y = np.array(self.sdm.y[j*batch_size:(j+1)*batch_size]))\n",
|
|
" if progress_callback is not None:\n",
|
|
" progress_callback()\n",
|
|
" pred, yt = self.test()\n",
|
|
" mean_squared_error = ((pred - yt)**2).mean(axis=0)\n",
|
|
" print(\"#\" + str(j) + \": loss: \", mean_squared_error)\n",
|
|
"\n",
|
|
" \n",
|
|
" # restore replaced fit functions:\n",
|
|
" for s in disabled_fit_steps:\n",
|
|
" named_steps[s].fit = disabled_fits[s]\n",
|
|
" named_steps[s].fit_transform = disabled_fit_transforms[s]\n",
|
|
" \n",
|
|
" for k in keras_batch_fitting_layer:\n",
|
|
" named_steps[k].fit = disabled_keras_fits[k]\n",
|
|
" \n",
|
|
" def test(self):\n",
|
|
" '''\n",
|
|
" @return: prediction:list, teacher:list\n",
|
|
" '''\n",
|
|
" if self.sdm.X is None:\n",
|
|
" self.sdm.create_train_test_split()\n",
|
|
" return self.pm.predict(self.sdm.Xt), self.sdm.yt\n"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"----\n",
|
|
"## Train"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"* when in notebook environment: run the stuff below:"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import __main__ as main\n",
|
|
"if not hasattr(main, '__file__'):\n",
|
|
" # we are in an interactive environment (probably in jupyter)\n",
|
|
" # load data:\n",
|
|
" \n",
|
|
" # setting n_kmeans_clusters to a value > 0 activates binarized labeling automatically! \n",
|
|
" # set to -1 to disable kmeans clustering and generating labels in plain sentiment space\n",
|
|
" \n",
|
|
" #n_kmeans_cluster = 5\n",
|
|
" n_kmeans_cluster = -1\n",
|
|
" sdm = sample_data_manager.generate_and_read(path=\"./data_en/\", n_top_emojis=20, file_range=range(1), n_kmeans_cluster=n_kmeans_cluster)\n",
|
|
" sdm.create_train_test_split()\n",
|
|
" #pm = pipeline_manager.create_keras_pipeline_with_vectorizer(vectorizer=TfidfVectorizer(stop_words='english'),\\n\",\n",
|
|
" # layers=[(10000, 'relu'),(5000, 'relu'),(2500, 'relu'),(y1[0].shape[0],None)], sdm=sdm)\\n\",\n",
|
|
" pm = pipeline_manager.create_keras_pipeline_with_vectorizer(vectorizer=TfidfVectorizer(stop_words='english'),\n",
|
|
" layers=[(2500, 'relu'),(sdm.y.shape[1],None)], sdm=sdm)\n",
|
|
" tr = trainer(sdm=sdm, pm=pm)\n",
|
|
" tr.fit(100)"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"----\n",
|
|
"## save classifier"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import __main__ as main\n",
|
|
"if not hasattr(main, '__file__'):\n",
|
|
" pm.save('custom_classifier')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"----\n",
|
|
"## Prediction\n",
|
|
"\n",
|
|
"* predict and save to `test.csv`"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import __main__ as main\n",
|
|
"if not hasattr(main, '__file__'):\n",
|
|
" pred, teacher = tr.test()\n",
|
|
" \n",
|
|
" display(pred)\n",
|
|
" display(teacher)\n",
|
|
" \n",
|
|
" print('prediction variance: ', np.linalg.norm(np.var(pred, axis=0)))\n",
|
|
" print('teacher variance: ', np.linalg.norm(np.var(teacher, axis=0)))\n",
|
|
" \n",
|
|
" # build a dataframe to visualize test results:\n",
|
|
" testlist = pd.DataFrame({'text': sdm.Xt, \n",
|
|
" 'teacher': sent2emoji(sdm.yt),\n",
|
|
" 'teacher_sentiment': sdm.yt.tolist(),\n",
|
|
" 'predict': sent2emoji(pred, custom_target_emojis=sdm.top_emojis),\n",
|
|
" 'predicted_sentiment': pred.tolist()})\n",
|
|
" # display:\n",
|
|
" display(testlist.head())\n",
|
|
" \n",
|
|
" # mean squared error:\n",
|
|
" teacher_sentiments = np.array([sample[1]['teacher_sentiment'] for sample in testlist.iterrows()])\n",
|
|
" predicted_sentiments = np.array([sample[1]['predicted_sentiment'] for sample in testlist.iterrows()])\n",
|
|
"\n",
|
|
" mean_squared_error = ((teacher_sentiments - predicted_sentiments)**2).mean(axis=0)\n",
|
|
" print(\"Mean Squared Error: \", mean_squared_error)\n",
|
|
" print(\"Variance teacher: \", np.var(teacher_sentiments, axis=0))\n",
|
|
" print(\"Variance prediction: \", np.var(predicted_sentiments, axis=0))\n",
|
|
" \n",
|
|
" # save to csv:\n",
|
|
" testlist.to_csv('test.csv')"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "markdown",
|
|
"metadata": {},
|
|
"source": [
|
|
"----\n",
|
|
"## Load classifier\n",
|
|
"\n",
|
|
"* loading classifier and show a test widget"
|
|
]
|
|
},
|
|
{
|
|
"cell_type": "code",
|
|
"execution_count": null,
|
|
"metadata": {},
|
|
"outputs": [],
|
|
"source": [
|
|
"import __main__ as main\n",
|
|
"if not hasattr(main, '__file__'):\n",
|
|
" try:\n",
|
|
" pm\n",
|
|
" except NameError:\n",
|
|
" pass\n",
|
|
" else:\n",
|
|
" del pm # delete existing pipeline manager if ther is one\n",
|
|
"\n",
|
|
" pm = pipeline_manager.load_pipeline_from_files( 'custom_classifier', ['keras_model'], ['vectorizer', 'keras_model'])\n",
|
|
" lookup_emojis = [#'😂',\n",
|
|
" '😭',\n",
|
|
" '😍',\n",
|
|
" '😩',\n",
|
|
" '😊',\n",
|
|
" '😘',\n",
|
|
" '🙏',\n",
|
|
" '🙌',\n",
|
|
" '😉',\n",
|
|
" '😁',\n",
|
|
" '😅',\n",
|
|
" '😎',\n",
|
|
" '😢',\n",
|
|
" '😒',\n",
|
|
" '😏',\n",
|
|
" '😌',\n",
|
|
" '😔',\n",
|
|
" '😋',\n",
|
|
" '😀',\n",
|
|
" '😤']\n",
|
|
" out = widgets.Output()\n",
|
|
"\n",
|
|
" t = widgets.Text()\n",
|
|
" b = widgets.Button(\n",
|
|
" description='get emoji',\n",
|
|
" disabled=False,\n",
|
|
" button_style='', # 'success', 'info', 'warning', 'danger' or ''\n",
|
|
" tooltip='Click me',\n",
|
|
" icon='check'\n",
|
|
" )\n",
|
|
"\n",
|
|
"\n",
|
|
"\n",
|
|
" def handle_submit(sender):\n",
|
|
" with out:\n",
|
|
" clear_output()\n",
|
|
" with out:\n",
|
|
" pred = pm.predict([t.value])\n",
|
|
"\n",
|
|
" display(Markdown(\"# Predicted Emoji \" + str(sent2emoji(pred, lookup_emojis)[0])))\n",
|
|
" display(Markdown(\"# Sentiment Vector: $$ \\pmatrix{\" + str(pred[0,0]) +\n",
|
|
" \"\\\\\\\\\" + str(pred[0,1]) + \"\\\\\\\\\" + str(pred[0,2]) + \"}$$\"))\n",
|
|
"\n",
|
|
" b.on_click(handle_submit)\n",
|
|
"\n",
|
|
" display(t)\n",
|
|
" display(widgets.VBox([b, out])) "
|
|
]
|
|
}
|
|
],
|
|
"metadata": {
|
|
"kernelspec": {
|
|
"display_name": "Python 3",
|
|
"language": "python",
|
|
"name": "python3"
|
|
},
|
|
"language_info": {
|
|
"codemirror_mode": {
|
|
"name": "ipython",
|
|
"version": 3
|
|
},
|
|
"file_extension": ".py",
|
|
"mimetype": "text/x-python",
|
|
"name": "python",
|
|
"nbconvert_exporter": "python",
|
|
"pygments_lexer": "ipython3",
|
|
"version": "3.6.5"
|
|
}
|
|
},
|
|
"nbformat": 4,
|
|
"nbformat_minor": 2
|
|
}
|