Similarity method make leave-one-out predictions.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pickle\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from gensim.models import KeyedVectors\n",
    "from itertools import product\n",
    "import seaborn as sns\n",
    "import matplotlib.pyplot as plt\n",
    "from scipy.stats import pearsonr\n",
    "import warnings\n",
    "from sklearn.model_selection import LeaveOneOut\n",
    "from sklearn.linear_model import LinearRegression\n",
    "from tqdm import tqdm\n",
    "\n",
    "warnings.filterwarnings(\"ignore\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [],
   "source": [
    "domains = ['brand', 'good', 'trait', 'food', 'occupation', 'risk', 'people']\n",
    "dims = [('sincere','exciting'),\n",
    "        ('hedonic','utilitarian'),\n",
    "        ('masculine','feminine'),\n",
    "        ('tasty','nutritious'),\n",
    "        ('significance','autonomy'),\n",
    "        ('dread-inducing','unknowable'),\n",
    "        ('warm', 'competent')]\n",
    "\n",
    "# domains = ['trait']\n",
    "# dims = [('masculine', 'feminine')]\n",
    "\n",
    "dim_names_flattened = [x for dim_pair in dims for x in dim_pair]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>0</th>\n",
       "      <th>1</th>\n",
       "      <th>2</th>\n",
       "      <th>3</th>\n",
       "      <th>4</th>\n",
       "      <th>5</th>\n",
       "      <th>6</th>\n",
       "      <th>7</th>\n",
       "      <th>8</th>\n",
       "      <th>9</th>\n",
       "      <th>...</th>\n",
       "      <th>290</th>\n",
       "      <th>291</th>\n",
       "      <th>292</th>\n",
       "      <th>293</th>\n",
       "      <th>294</th>\n",
       "      <th>295</th>\n",
       "      <th>296</th>\n",
       "      <th>297</th>\n",
       "      <th>298</th>\n",
       "      <th>299</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>sincere</th>\n",
       "      <td>-0.090576</td>\n",
       "      <td>0.056602</td>\n",
       "      <td>-0.059738</td>\n",
       "      <td>0.230011</td>\n",
       "      <td>0.081116</td>\n",
       "      <td>0.090279</td>\n",
       "      <td>0.228943</td>\n",
       "      <td>-0.005371</td>\n",
       "      <td>-0.083893</td>\n",
       "      <td>-0.082916</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.096397</td>\n",
       "      <td>-0.038284</td>\n",
       "      <td>0.026642</td>\n",
       "      <td>-0.122517</td>\n",
       "      <td>-0.019836</td>\n",
       "      <td>0.098328</td>\n",
       "      <td>0.054337</td>\n",
       "      <td>-0.051895</td>\n",
       "      <td>0.130005</td>\n",
       "      <td>-0.097415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>exciting</th>\n",
       "      <td>-0.134979</td>\n",
       "      <td>-0.131195</td>\n",
       "      <td>-0.031158</td>\n",
       "      <td>-0.109253</td>\n",
       "      <td>0.154205</td>\n",
       "      <td>-0.242981</td>\n",
       "      <td>-0.252869</td>\n",
       "      <td>0.102875</td>\n",
       "      <td>-0.137268</td>\n",
       "      <td>-0.061630</td>\n",
       "      <td>...</td>\n",
       "      <td>0.025345</td>\n",
       "      <td>-0.081116</td>\n",
       "      <td>-0.119629</td>\n",
       "      <td>-0.248619</td>\n",
       "      <td>0.087677</td>\n",
       "      <td>-0.021038</td>\n",
       "      <td>-0.006592</td>\n",
       "      <td>-0.044891</td>\n",
       "      <td>0.047729</td>\n",
       "      <td>-0.097046</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>hedonic</th>\n",
       "      <td>-0.094686</td>\n",
       "      <td>-0.138184</td>\n",
       "      <td>-0.052444</td>\n",
       "      <td>-0.001607</td>\n",
       "      <td>0.130692</td>\n",
       "      <td>-0.265666</td>\n",
       "      <td>-0.163249</td>\n",
       "      <td>0.034688</td>\n",
       "      <td>-0.263947</td>\n",
       "      <td>0.026693</td>\n",
       "      <td>...</td>\n",
       "      <td>0.022400</td>\n",
       "      <td>-0.000488</td>\n",
       "      <td>-0.030711</td>\n",
       "      <td>-0.190470</td>\n",
       "      <td>0.060465</td>\n",
       "      <td>-0.013041</td>\n",
       "      <td>0.049377</td>\n",
       "      <td>-0.024394</td>\n",
       "      <td>0.098256</td>\n",
       "      <td>-0.171712</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>utilitarian</th>\n",
       "      <td>-0.153661</td>\n",
       "      <td>-0.243525</td>\n",
       "      <td>-0.179016</td>\n",
       "      <td>-0.005483</td>\n",
       "      <td>0.156576</td>\n",
       "      <td>-0.251504</td>\n",
       "      <td>-0.033610</td>\n",
       "      <td>0.030248</td>\n",
       "      <td>-0.071508</td>\n",
       "      <td>-0.213359</td>\n",
       "      <td>...</td>\n",
       "      <td>-0.204427</td>\n",
       "      <td>-0.147685</td>\n",
       "      <td>0.080009</td>\n",
       "      <td>-0.008789</td>\n",
       "      <td>0.112905</td>\n",
       "      <td>0.175964</td>\n",
       "      <td>0.139882</td>\n",
       "      <td>-0.102478</td>\n",
       "      <td>0.159098</td>\n",
       "      <td>-0.065104</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>masculine</th>\n",
       "      <td>-0.103109</td>\n",
       "      <td>0.088967</td>\n",
       "      <td>0.124990</td>\n",
       "      <td>-0.216858</td>\n",
       "      <td>0.165080</td>\n",
       "      <td>-0.187937</td>\n",
       "      <td>-0.153849</td>\n",
       "      <td>0.115448</td>\n",
       "      <td>-0.010559</td>\n",
       "      <td>-0.054955</td>\n",
       "      <td>...</td>\n",
       "      <td>0.028158</td>\n",
       "      <td>-0.133870</td>\n",
       "      <td>0.070394</td>\n",
       "      <td>-0.058451</td>\n",
       "      <td>0.039907</td>\n",
       "      <td>0.012098</td>\n",
       "      <td>-0.149007</td>\n",
       "      <td>-0.273158</td>\n",
       "      <td>0.033691</td>\n",
       "      <td>0.016154</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 300 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "                    0         1         2         3         4         5  \\\n",
       "sincere     -0.090576  0.056602 -0.059738  0.230011  0.081116  0.090279   \n",
       "exciting    -0.134979 -0.131195 -0.031158 -0.109253  0.154205 -0.242981   \n",
       "hedonic     -0.094686 -0.138184 -0.052444 -0.001607  0.130692 -0.265666   \n",
       "utilitarian -0.153661 -0.243525 -0.179016 -0.005483  0.156576 -0.251504   \n",
       "masculine   -0.103109  0.088967  0.124990 -0.216858  0.165080 -0.187937   \n",
       "\n",
       "                    6         7         8         9    ...          290  \\\n",
       "sincere      0.228943 -0.005371 -0.083893 -0.082916    ...    -0.096397   \n",
       "exciting    -0.252869  0.102875 -0.137268 -0.061630    ...     0.025345   \n",
       "hedonic     -0.163249  0.034688 -0.263947  0.026693    ...     0.022400   \n",
       "utilitarian -0.033610  0.030248 -0.071508 -0.213359    ...    -0.204427   \n",
       "masculine   -0.153849  0.115448 -0.010559 -0.054955    ...     0.028158   \n",
       "\n",
       "                  291       292       293       294       295       296  \\\n",
       "sincere     -0.038284  0.026642 -0.122517 -0.019836  0.098328  0.054337   \n",
       "exciting    -0.081116 -0.119629 -0.248619  0.087677 -0.021038 -0.006592   \n",
       "hedonic     -0.000488 -0.030711 -0.190470  0.060465 -0.013041  0.049377   \n",
       "utilitarian -0.147685  0.080009 -0.008789  0.112905  0.175964  0.139882   \n",
       "masculine   -0.133870  0.070394 -0.058451  0.039907  0.012098 -0.149007   \n",
       "\n",
       "                  297       298       299  \n",
       "sincere     -0.051895  0.130005 -0.097415  \n",
       "exciting    -0.044891  0.047729 -0.097046  \n",
       "hedonic     -0.024394  0.098256 -0.171712  \n",
       "utilitarian -0.102478  0.159098 -0.065104  \n",
       "masculine   -0.273158  0.033691  0.016154  \n",
       "\n",
       "[5 rows x 300 columns]"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# dim_vecs_df = pd.read_csv('data/word2vec_similarity_dim_vecs.csv', index_col=0)\n",
    "dim_vecs_df = pd.read_csv('data/word2vec_similarity_alt_gender_dim_vecs.csv', index_col=0)\n",
    "# dim_vecs_df = pd.read_csv('data/paragram_similarity_dim_vecs.csv', index_col=0)\n",
    "dim_vecs_df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"data/dict_of_Xs.pickle\", \"rb\") as handle:\n",
    "    dict_of_Xs = pickle.load(handle)\n",
    "\n",
    "# with open(\"data/dict_of_Xs_paragram.pickle\", \"rb\") as handle:\n",
    "#     dict_of_Xs = pickle.load(handle)\n",
    "\n",
    "with open(\"data/dict_of_ys.pickle\", \"rb\") as handle:\n",
    "    dict_of_ys = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "dict_keys(['brand', 'good', 'trait', 'food', 'occupation', 'risk', 'people'])"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dict_of_Xs.keys()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Modeling means"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# correlation_results = []\n",
    "\n",
    "regr = LinearRegression()\n",
    "\n",
    "for domain, dim_pair in zip(domains, dims):\n",
    "    X = dict_of_Xs[domain]\n",
    "    for dim in dim_pair:\n",
    "        dim_vec = dim_vecs_df.loc[dim].values\n",
    "        dot_products = X.applymap(float).dot(dim_vec).to_frame()\n",
    "        y = dict_of_ys[dim]\n",
    "        y.name = f'actual_{dim}'\n",
    "\n",
    "        y_preds = np.zeros(shape=len(y))\n",
    "        loo = LeaveOneOut()\n",
    "        for train_index, test_index in loo.split(dot_products):\n",
    "            dot_products_train, dot_products_test = dot_products.iloc[train_index], dot_products.iloc[test_index]\n",
    "            y_train, y_test = y[train_index], y[test_index]\n",
    "            regr.fit(X=dot_products_train, y=y_train)\n",
    "            y_pred = regr.predict(X=dot_products_test)\n",
    "            y_preds[test_index] = y_pred\n",
    "            \n",
    "        preds_and_actual =  pd.DataFrame({f'Actual {dim}':y, f'Predicted {dim}':y_preds}, index=dict_of_ys[dim].index)\n",
    "        preds_and_actual.to_csv(f'results/grand_method_actual_vs_pred/{dim}.csv')                      \n",
    "#         preds_and_actual.to_csv(f'results/grand_method_actual_vs_pred_paragram/{dim}.csv')                      \n",
    "#         dot_products = pd.Series(dot_products, name=f'predicted_{dim}')\n",
    "\n",
    "#         sns.jointplot(x=dot_products, y=y)\n",
    "#         plt.savefig(f'figures/pred_v_actual_scatterplots_grand_et_al/{dim}.png')\n",
    "#         plt.show()\n",
    "        \n",
    "#         r, p = pearsonr(dot_products, y)\n",
    "#         correlation_results.append((r,p, r**2))\n",
    "        \n",
    "# correlation_results = pd.DataFrame(correlation_results, index=dim_names_flattened, columns=['r','p','rsquared'])\n",
    "# correlation_results.to_csv('results/grand_et_al_method.csv', float_format='%.4f')"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Modeling individuals"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"data/dict_of_individual_ys.pickle\", \"rb\") as handle:\n",
    "    dict_of_individual_ys = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [],
   "source": [
    "regr = LinearRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [],
   "source": [
    "dict_of_individual_similarity_y_preds = dict()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "7it [01:38, 14.06s/it]\n"
     ]
    }
   ],
   "source": [
    "for domain, dim_pair in tqdm(zip(domains, dims)):\n",
    "    X = dict_of_Xs[domain]\n",
    "    for dim in dim_pair:\n",
    "        dim_vec = dim_vecs_df.loc[dim].values\n",
    "        dot_products = X.applymap(float).dot(dim_vec).to_frame()\n",
    "        \n",
    "        all_individual_preds = np.zeros(shape=dict_of_individual_ys[dim].shape)\n",
    "        for individual_idx, y in enumerate(dict_of_individual_ys[dim].values):\n",
    "            y_preds = np.zeros(shape=len(y))\n",
    "            loo = LeaveOneOut()\n",
    "            for train_index, test_index in loo.split(dot_products):\n",
    "                X_train = dot_products.iloc[train_index] \n",
    "                X_test  = dot_products.iloc[test_index]\n",
    "                y_train = y[train_index].astype(float)\n",
    "                \n",
    "                y_notna_idxs = ~np.isnan(y_train)\n",
    "                y_train = y_train[y_notna_idxs]\n",
    "                X_train = X_train[y_notna_idxs]\n",
    "                \n",
    "                regr.fit(X=X_train, y=y_train)\n",
    "                y_pred = regr.predict(X=X_test)[0]\n",
    "                y_preds[test_index] = y_pred\n",
    "            all_individual_preds[individual_idx] = y_preds\n",
    "        individual_predictions = pd.DataFrame(all_individual_preds, \n",
    "                                              columns=dict_of_individual_ys[dim].columns)\n",
    "        dict_of_individual_similarity_y_preds[dim] = individual_predictions"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [],
   "source": [
    "with open(\"results/dict_of_individual_similarity_y_preds.pickle\", \"wb\") as handle:\n",
    "# with open(\"results/dict_of_individual_similarity_y_preds_paragram.pickle\", \"wb\") as handle:\n",
    "    pickle.dump(dict_of_individual_similarity_y_preds, handle, protocol=pickle.HIGHEST_PROTOCOL)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Compute and visualize individual-level actual vs pred correlations"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "if 'dict_of_individual_similarity_y_preds' not in globals():\n",
    "    with open(\"results/dict_of_individual_similarity_y_preds.pickle\", \"rb\") as handle:\n",
    "#     with open(\"results/dict_of_individual_similarity_y_preds_paragram.pickle\", \"rb\") as handle:\n",
    "        dict_of_individual_similarity_y_preds = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_participant_rs = []\n",
    "dims = dict_of_individual_ys.keys()\n",
    "for dim in dims:\n",
    "    df_ys      = dict_of_individual_ys[dim]\n",
    "    df_y_preds = dict_of_individual_similarity_y_preds[dim]\n",
    "    dim_rs = []\n",
    "    for ys, y_preds in zip(df_ys.values, df_y_preds.values):\n",
    "        y_not_na_idx = ~np.isnan(ys.astype(float))\n",
    "\n",
    "        ys_not_na      = ys[y_not_na_idx]\n",
    "        y_preds_not_na = y_preds[y_not_na_idx]\n",
    "        \n",
    "        r, _ = pearsonr(ys_not_na, y_preds_not_na)\n",
    "        dim_rs.append(r)\n",
    "    all_participant_rs.append(dim_rs)\n",
    "all_participant_rs_df = pd.DataFrame(all_participant_rs, index=dims).T"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_participant_rs_df_stacked = all_participant_rs_df.stack().reset_index()\n",
    "all_participant_rs_df_stacked.columns = ['participant_number', 'Judgment Dimension', 'Correlation of Pred vs Actual']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "participant_number               17.357143\n",
       "Correlation of Pred vs Actual     0.211052\n",
       "dtype: float64"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "all_participant_rs_df_stacked.groupby('Judgment Dimension').mean().abs().mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [],
   "source": [
    "all_participant_rs_df_stacked.to_csv('results/individual-level_similarity_model_results.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x576 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.figure(figsize=(15,8))\n",
    "plt.xticks(rotation=45);\n",
    "sns.barplot(data=all_participant_rs_df_stacked, x='Judgment Dimension', y='Correlation of Pred vs Actual', color='gray');\n",
    "plt.savefig('figures/individual_level_similarity_pred_v_actual_correlation_barplots.png')\n",
    "# plt.savefig('figures/individual_level_similarity_pred_v_actual_correlation_barplots_paragram.png')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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
}