method chain & func added
This commit is contained in:
almsam
parent
cf131c39bb
commit
c18048c125
|
|
@ -875,7 +875,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 9,
|
"execution_count": 77,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1080,7 +1080,7 @@
|
||||||
"[5 rows x 29 columns]"
|
"[5 rows x 29 columns]"
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
"execution_count": 9,
|
"execution_count": 77,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"output_type": "execute_result"
|
"output_type": "execute_result"
|
||||||
}
|
}
|
||||||
|
|
@ -1096,47 +1096,37 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 27,
|
"execution_count": 82,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
"name": "stdout",
|
"name": "stdout",
|
||||||
"output_type": "stream",
|
"output_type": "stream",
|
||||||
"text": [
|
"text": [
|
||||||
" zip pride parade index gay bars index ZIP \\\n",
|
" pride parade index gay bars index lat long\n",
|
||||||
"count 2328.000000 2328.000000 2328.000000 2328.000000\n",
|
"count 2328.000000 2328.000000 2328.000000 2328.000000\n",
|
||||||
"mean 48616.478522 0.017612 0.118127 48616.478522 \n",
|
"mean 0.017612 0.118127 38.044304 -91.221236\n",
|
||||||
"std 35481.240641 0.131563 0.861863 35481.240641 \n",
|
"std 0.131563 0.861863 5.148365 18.533499\n",
|
||||||
"min 1730.000000 0.000000 0.000000 1730.000000 \n",
|
"min 0.000000 0.000000 25.572213 -123.118977\n",
|
||||||
"25% 11362.750000 0.000000 0.000000 11362.750000 \n",
|
"25% 0.000000 0.000000 34.021932 -105.049099\n",
|
||||||
"50% 46351.000000 0.000000 0.000000 46351.000000 \n",
|
"50% 0.000000 0.000000 39.899977 -87.494097\n",
|
||||||
"75% 80234.250000 0.000000 0.000000 80234.250000 \n",
|
"75% 0.000000 0.000000 40.912413 -74.288743\n",
|
||||||
"max 98686.000000 1.000000 17.000000 98686.000000 \n",
|
"max 1.000000 17.000000 47.916786 -70.758184\n",
|
||||||
"\n",
|
|
||||||
" lat long \n",
|
|
||||||
"count 2328.000000 2328.000000 \n",
|
|
||||||
"mean 38.044304 -91.221236 \n",
|
|
||||||
"std 5.148365 18.533499 \n",
|
|
||||||
"min 25.572213 -123.118977 \n",
|
|
||||||
"25% 34.021932 -105.049099 \n",
|
|
||||||
"50% 39.899977 -87.494097 \n",
|
|
||||||
"75% 40.912413 -74.288743 \n",
|
|
||||||
"max 47.916786 -70.758184 \n",
|
|
||||||
"------------------------------------------------------------------------\n",
|
"------------------------------------------------------------------------\n",
|
||||||
" zip pride parade index gay bars index ZIP lat long\n",
|
" pride parade index gay bars index lat long\n",
|
||||||
"0 90069 1 15 90069 34.093828 -118.381697\n",
|
"0 1 15 34.093828 -118.381697\n",
|
||||||
"1 94114 0 17 94114 37.758057 -122.435410\n",
|
"1 0 17 37.758057 -122.435410\n",
|
||||||
"2 10011 1 5 10011 40.742039 -74.000620\n",
|
"2 1 5 40.742039 -74.000620\n",
|
||||||
"3 10014 1 10 10014 40.734012 -74.006746\n",
|
"3 1 10 40.734012 -74.006746\n",
|
||||||
"4 94103 1 9 94103 37.773134 -122.411167\n",
|
"4 1 9 37.773134 -122.411167\n",
|
||||||
"... ... ... ... ... ... ...\n",
|
"... ... ... ... ...\n",
|
||||||
"2323 97208 0 0 97208 45.528666 -122.678981\n",
|
"2323 0 0 45.528666 -122.678981\n",
|
||||||
"2324 98154 0 0 98154 47.606211 -122.333792\n",
|
"2324 0 0 47.606211 -122.333792\n",
|
||||||
"2325 98158 0 0 98158 47.449678 -122.307657\n",
|
"2325 0 0 47.449678 -122.307657\n",
|
||||||
"2326 98174 0 0 98174 47.604569 -122.335359\n",
|
"2326 0 0 47.604569 -122.335359\n",
|
||||||
"2327 98195 0 0 98195 47.649339 -122.310294\n",
|
"2327 0 0 47.649339 -122.310294\n",
|
||||||
"\n",
|
"\n",
|
||||||
"[2328 rows x 6 columns]\n"
|
"[2328 rows x 4 columns]\n"
|
||||||
]
|
]
|
||||||
}
|
}
|
||||||
],
|
],
|
||||||
|
|
@ -1144,16 +1134,38 @@
|
||||||
"#wrangle gay\n",
|
"#wrangle gay\n",
|
||||||
"# gaydf = pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['Parade_Weight'], gaybourhoods['Bars_Weight']))\n",
|
"# gaydf = pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['Parade_Weight'], gaybourhoods['Bars_Weight']))\n",
|
||||||
"\n",
|
"\n",
|
||||||
"gaydf = pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['ParadeFlag'], gaybourhoods['CountBars']))\n",
|
|
||||||
"\n",
|
|
||||||
"gaydf.columns=(('zip', 'pride parade index', 'gay bars index'))\n",
|
|
||||||
"\n",
|
"\n",
|
||||||
"# print(gaydf.describe())\n",
|
"# print(gaydf.describe())\n",
|
||||||
"# print(gaydf)\n",
|
"# print(gaydf)\n",
|
||||||
"\n",
|
"\n",
|
||||||
"cords = pd.read_csv(\"../data/raw/zip_lat_long.csv\")\n",
|
"cords = pd.read_csv(\"../data/raw/zip_lat_long.csv\")\n",
|
||||||
"\n",
|
"\n",
|
||||||
"gaydf = gaydf.merge(cords, left_on=\"zip\", right_on=\"ZIP\")\n",
|
"# gaydf = pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['ParadeFlag'], gaybourhoods['CountBars']))\n",
|
||||||
|
"# gaydf.columns=(('zip', 'pride parade index', 'gay bars index'))\n",
|
||||||
|
"# gaydf = gaydf.merge(cords, left_on=\"zip\", right_on=\"ZIP\")\n",
|
||||||
|
"\n",
|
||||||
|
"\n",
|
||||||
|
"def loadprideindexes(secdf):\n",
|
||||||
|
" gaydf = pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['ParadeFlag'], gaybourhoods['CountBars']))\n",
|
||||||
|
" gaydf.columns=(('zip', 'pride parade index', 'gay bars index'))\n",
|
||||||
|
" gaydf = gaydf.merge(secdf, left_on=\"zip\", right_on=\"ZIP\")\n",
|
||||||
|
" \n",
|
||||||
|
" return gaydf\n",
|
||||||
|
" \n",
|
||||||
|
" # gaydf = (\n",
|
||||||
|
" # pd.DataFrame(zip(gaybourhoods['GEOID10'], gaybourhoods['ParadeFlag'], gaybourhoods['CountBars']))\n",
|
||||||
|
" # .columns=(('zip', 'pride parade index', 'gay bars index'))\n",
|
||||||
|
" # # .merge(cords, left_on=\"zip\", right_on=\"ZIP\")\n",
|
||||||
|
" # )\n",
|
||||||
|
"\n",
|
||||||
|
"gaydf = loadprideindexes(cords)\n",
|
||||||
|
"\n",
|
||||||
|
"def notzip(dat):\n",
|
||||||
|
" del dat['zip']\n",
|
||||||
|
" del dat['ZIP']\n",
|
||||||
|
"\n",
|
||||||
|
"notzip(gaydf)\n",
|
||||||
|
"\n",
|
||||||
"\n",
|
"\n",
|
||||||
"# // unneded was already filtered out\n",
|
"# // unneded was already filtered out\n",
|
||||||
"\n",
|
"\n",
|
||||||
|
|
@ -1179,7 +1191,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 28,
|
"execution_count": 53,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [],
|
"outputs": [],
|
||||||
"source": [
|
"source": [
|
||||||
|
|
@ -1188,7 +1200,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 29,
|
"execution_count": 54,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1219,7 +1231,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 34,
|
"execution_count": 55,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1252,7 +1264,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 35,
|
"execution_count": 56,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1280,7 +1292,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 31,
|
"execution_count": 57,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1304,10 +1316,8 @@
|
||||||
" <thead>\n",
|
" <thead>\n",
|
||||||
" <tr style=\"text-align: right;\">\n",
|
" <tr style=\"text-align: right;\">\n",
|
||||||
" <th></th>\n",
|
" <th></th>\n",
|
||||||
" <th>zip</th>\n",
|
|
||||||
" <th>pride parade index</th>\n",
|
" <th>pride parade index</th>\n",
|
||||||
" <th>gay bars index</th>\n",
|
" <th>gay bars index</th>\n",
|
||||||
" <th>ZIP</th>\n",
|
|
||||||
" <th>lat</th>\n",
|
" <th>lat</th>\n",
|
||||||
" <th>long</th>\n",
|
" <th>long</th>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
|
|
@ -1319,69 +1329,53 @@
|
||||||
" <td>2328.000000</td>\n",
|
" <td>2328.000000</td>\n",
|
||||||
" <td>2328.000000</td>\n",
|
" <td>2328.000000</td>\n",
|
||||||
" <td>2328.000000</td>\n",
|
" <td>2328.000000</td>\n",
|
||||||
" <td>2328.000000</td>\n",
|
|
||||||
" <td>2328.000000</td>\n",
|
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>mean</th>\n",
|
" <th>mean</th>\n",
|
||||||
" <td>48616.478522</td>\n",
|
|
||||||
" <td>0.017612</td>\n",
|
" <td>0.017612</td>\n",
|
||||||
" <td>0.118127</td>\n",
|
" <td>0.118127</td>\n",
|
||||||
" <td>48616.478522</td>\n",
|
|
||||||
" <td>38.044304</td>\n",
|
" <td>38.044304</td>\n",
|
||||||
" <td>-91.221236</td>\n",
|
" <td>-91.221236</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>std</th>\n",
|
" <th>std</th>\n",
|
||||||
" <td>35481.240641</td>\n",
|
|
||||||
" <td>0.131563</td>\n",
|
" <td>0.131563</td>\n",
|
||||||
" <td>0.861863</td>\n",
|
" <td>0.861863</td>\n",
|
||||||
" <td>35481.240641</td>\n",
|
|
||||||
" <td>5.148365</td>\n",
|
" <td>5.148365</td>\n",
|
||||||
" <td>18.533499</td>\n",
|
" <td>18.533499</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>min</th>\n",
|
" <th>min</th>\n",
|
||||||
" <td>1730.000000</td>\n",
|
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>1730.000000</td>\n",
|
|
||||||
" <td>25.572213</td>\n",
|
" <td>25.572213</td>\n",
|
||||||
" <td>-123.118977</td>\n",
|
" <td>-123.118977</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>25%</th>\n",
|
" <th>25%</th>\n",
|
||||||
" <td>11362.750000</td>\n",
|
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>11362.750000</td>\n",
|
|
||||||
" <td>34.021932</td>\n",
|
" <td>34.021932</td>\n",
|
||||||
" <td>-105.049099</td>\n",
|
" <td>-105.049099</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>50%</th>\n",
|
" <th>50%</th>\n",
|
||||||
" <td>46351.000000</td>\n",
|
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>46351.000000</td>\n",
|
|
||||||
" <td>39.899977</td>\n",
|
" <td>39.899977</td>\n",
|
||||||
" <td>-87.494097</td>\n",
|
" <td>-87.494097</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>75%</th>\n",
|
" <th>75%</th>\n",
|
||||||
" <td>80234.250000</td>\n",
|
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>0.000000</td>\n",
|
" <td>0.000000</td>\n",
|
||||||
" <td>80234.250000</td>\n",
|
|
||||||
" <td>40.912413</td>\n",
|
" <td>40.912413</td>\n",
|
||||||
" <td>-74.288743</td>\n",
|
" <td>-74.288743</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
" <tr>\n",
|
" <tr>\n",
|
||||||
" <th>max</th>\n",
|
" <th>max</th>\n",
|
||||||
" <td>98686.000000</td>\n",
|
|
||||||
" <td>1.000000</td>\n",
|
" <td>1.000000</td>\n",
|
||||||
" <td>17.000000</td>\n",
|
" <td>17.000000</td>\n",
|
||||||
" <td>98686.000000</td>\n",
|
|
||||||
" <td>47.916786</td>\n",
|
" <td>47.916786</td>\n",
|
||||||
" <td>-70.758184</td>\n",
|
" <td>-70.758184</td>\n",
|
||||||
" </tr>\n",
|
" </tr>\n",
|
||||||
|
|
@ -1390,28 +1384,18 @@
|
||||||
"</div>"
|
"</div>"
|
||||||
],
|
],
|
||||||
"text/plain": [
|
"text/plain": [
|
||||||
" zip pride parade index gay bars index ZIP \\\n",
|
" pride parade index gay bars index lat long\n",
|
||||||
"count 2328.000000 2328.000000 2328.000000 2328.000000\n",
|
"count 2328.000000 2328.000000 2328.000000 2328.000000\n",
|
||||||
"mean 48616.478522 0.017612 0.118127 48616.478522 \n",
|
"mean 0.017612 0.118127 38.044304 -91.221236\n",
|
||||||
"std 35481.240641 0.131563 0.861863 35481.240641 \n",
|
"std 0.131563 0.861863 5.148365 18.533499\n",
|
||||||
"min 1730.000000 0.000000 0.000000 1730.000000 \n",
|
"min 0.000000 0.000000 25.572213 -123.118977\n",
|
||||||
"25% 11362.750000 0.000000 0.000000 11362.750000 \n",
|
"25% 0.000000 0.000000 34.021932 -105.049099\n",
|
||||||
"50% 46351.000000 0.000000 0.000000 46351.000000 \n",
|
"50% 0.000000 0.000000 39.899977 -87.494097\n",
|
||||||
"75% 80234.250000 0.000000 0.000000 80234.250000 \n",
|
"75% 0.000000 0.000000 40.912413 -74.288743\n",
|
||||||
"max 98686.000000 1.000000 17.000000 98686.000000 \n",
|
"max 1.000000 17.000000 47.916786 -70.758184"
|
||||||
"\n",
|
|
||||||
" lat long \n",
|
|
||||||
"count 2328.000000 2328.000000 \n",
|
|
||||||
"mean 38.044304 -91.221236 \n",
|
|
||||||
"std 5.148365 18.533499 \n",
|
|
||||||
"min 25.572213 -123.118977 \n",
|
|
||||||
"25% 34.021932 -105.049099 \n",
|
|
||||||
"50% 39.899977 -87.494097 \n",
|
|
||||||
"75% 40.912413 -74.288743 \n",
|
|
||||||
"max 47.916786 -70.758184 "
|
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
"execution_count": 31,
|
"execution_count": 57,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"output_type": "execute_result"
|
"output_type": "execute_result"
|
||||||
}
|
}
|
||||||
|
|
@ -1424,16 +1408,16 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 32,
|
"execution_count": 58,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
"data": {
|
"data": {
|
||||||
"text/plain": [
|
"text/plain": [
|
||||||
"<seaborn.axisgrid.PairGrid at 0x2aec9f50910>"
|
"<seaborn.axisgrid.PairGrid at 0x2aecc151110>"
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
"execution_count": 32,
|
"execution_count": 58,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"output_type": "execute_result"
|
"output_type": "execute_result"
|
||||||
},
|
},
|
||||||
|
|
@ -1449,23 +1433,23 @@
|
||||||
}
|
}
|
||||||
],
|
],
|
||||||
"source": [
|
"source": [
|
||||||
"del gaydfhybridplot2['zip']; del gaydfhybridplot2['ZIP']\n",
|
"# del gaydfhybridplot2['zip']; del gaydfhybridplot2['ZIP']\n",
|
||||||
"\n",
|
"\n",
|
||||||
"sns.pairplot(gaydfhybridplot2)"
|
"sns.pairplot(gaydfhybridplot2)"
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 38,
|
"execution_count": 59,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
"data": {
|
"data": {
|
||||||
"text/plain": [
|
"text/plain": [
|
||||||
"<seaborn.axisgrid.PairGrid at 0x2aecadf1110>"
|
"<seaborn.axisgrid.PairGrid at 0x2aecd0e1750>"
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
"execution_count": 38,
|
"execution_count": 59,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"output_type": "execute_result"
|
"output_type": "execute_result"
|
||||||
},
|
},
|
||||||
|
|
@ -1498,16 +1482,16 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 42,
|
"execution_count": 60,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
"data": {
|
"data": {
|
||||||
"text/plain": [
|
"text/plain": [
|
||||||
"<seaborn.axisgrid.PairGrid at 0x2aebe5ed150>"
|
"<seaborn.axisgrid.PairGrid at 0x2aebcea1e50>"
|
||||||
]
|
]
|
||||||
},
|
},
|
||||||
"execution_count": 42,
|
"execution_count": 60,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"output_type": "execute_result"
|
"output_type": "execute_result"
|
||||||
},
|
},
|
||||||
|
|
@ -1531,7 +1515,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 47,
|
"execution_count": 61,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
@ -1558,7 +1542,7 @@
|
||||||
},
|
},
|
||||||
{
|
{
|
||||||
"cell_type": "code",
|
"cell_type": "code",
|
||||||
"execution_count": 48,
|
"execution_count": 62,
|
||||||
"metadata": {},
|
"metadata": {},
|
||||||
"outputs": [
|
"outputs": [
|
||||||
{
|
{
|
||||||
|
|
|
||||||
File diff suppressed because it is too large
Load Diff
Loading…
Reference in New Issue