added ft work

notebooks/.ipynb_checkpoints/make_bert_embeddings-checkpoint.ipynb

@@ -0,0 +1,91 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "36ccfe91-a69d-4077-8c03-32f97e3ac02a",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import os\n",
+    "os.environ['TRANSFORMERS_CACHE'] = '/Users/baga_nuhkadiev/.cache/huggingface'"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "d50e667e-3209-42f9-8612-6e64405c5107",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "from transformers import BertTokenizer, BertModel\n",
+    "\n",
+    "tokenizer = BertTokenizer.from_pretrained('bert-base-uncased')\n",
+    "model = BertModel.from_pretrained('bert-base-uncased')\n",
+    "\n",
+    "def get_bert_embeddings(text):\n",
+    "    inputs = tokenizer(text, return_tensors='pt', truncation=True, padding=True, max_length=512)\n",
+    "    outputs = model(**inputs)\n",
+    "    # Get the embeddings from the last hidden state\n",
+    "    embeddings = outputs.last_hidden_state\n",
+    "    # Pool the embeddings (use mean pooling for simplicity)\n",
+    "    pooled_embeddings = torch.mean(embeddings, dim=1)\n",
+    "    return pooled_embeddings.detach().numpy()\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "3df9f25b-1dbb-496b-b037-ffe692ef3831",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "headlines = data['Text'].values\n",
+    "embeddings = [get_bert_embeddings(headline) for headline in headlines]\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "7b42119a-45a4-4bfd-9013-e7005ebd18f1",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "bert_embeddings = np.array(embeddings)\n",
+    "bert_embeddings.shape"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "c0489af6-64dc-4386-a1e9-049f2dcee9ac",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "with open('../bert_embeddings.npy', 'wb') as f:\n",
+    "    np.save(f, bert_embeddings)\n"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.8.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

notebooks/.ipynb_checkpoints/make_fasttext_embeddings-checkpoint.ipynb

@@ -0,0 +1,104 @@
+{
+ "cells": [
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "da379302-b109-494a-a5b5-b97b52ae6e77",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import torch\n",
+    "import torch.nn as nn\n",
+    "import numpy as np\n",
+    "import pandas as pd\n",
+    "from torch.utils.data import TensorDataset, DataLoader\n",
+    "from nltk import WordPunctTokenizer\n",
+    "from nltk.corpus import stopwords\n",
+    "import nltk\n",
+    "import math"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "23567fa5-f680-4180-973a-416f7fa9e8cd",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "t_data = pd.DataFrame()\n",
+    "\n",
+    "combined_news_djia = pd.read_csv('../data/Combined_News_DJIA.csv')\n",
+    "combined_news_djia['Top1'] = combined_news_djia['Top1'].apply(lambda x: x[2:-1] if x[0]=='b' else x)\n",
+    "combined_news_djia['Top2'] = combined_news_djia['Top2'].apply(lambda x: x[2:-1] if x[0]=='b' else x)\n",
+    "\n",
+    "t_data['Text'] = combined_news_djia['Top1'] + \" \" + combined_news_djia['Top2']\n",
+    "t_data['Date'] = combined_news_djia['Date']\n",
+    "t_data.set_index('Date')\n",
+    "\n",
+    "nltk.download('stopwords')\n",
+    "tokenizer = WordPunctTokenizer()\n",
+    "stop_words = set(stopwords.words('english'))\n",
+    "\n",
+    "def process_headline(x):\n",
+    "    return \" \".join([w.lower() for w in tokenizer.tokenize(x) if not w.lower() in stop_words])\n",
+    "\n",
+    "t_data['Text'] = t_data['Text'].apply(process_headline)\n",
+    "t_data"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "35019d74-fd71-44b2-aa59-311f4cc1b3dc",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import fasttext.util\n",
+    "\n",
+    "# Download FastText word vectors\n",
+    "# fasttext.util.download_model('en', if_exists='ignore')  # Download English language embeddings\n",
+    "ft = fasttext.load_model('../../NLPstockPredictions/cc.en.300.bin')  # Load the downloaded model\n",
+    "\n",
+    "def get_embeddings(data):\n",
+    "    combo = []\n",
+    "    for row in data.values:\n",
+    "        news_embedding = np.mean([ft.get_word_vector(word) for word in row[0].split()], axis=0)\n",
+    "        combo.append(news_embedding)\n",
+    "    return np.array(combo), data.values[:, 1]\n"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "2a214078-569d-4ef4-9fd2-ad8a7b114557",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "t_data = get_embeddings(t_data)\n",
+    "with open('../fasttext_embeddings.npy', 'wb') as f:\n",
+    "    np.save(f, t_data[0])"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3 (ipykernel)",
+   "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": "3.8.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}