Spaces:
Runtime error
Runtime error
Commit
•
0928f44
1
Parent(s):
b468dba
Upload 1_data_processing.ipynb
Browse files- 1_data_processing.ipynb +356 -0
1_data_processing.ipynb
ADDED
@@ -0,0 +1,356 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
{
|
2 |
+
"cells": [
|
3 |
+
{
|
4 |
+
"cell_type": "code",
|
5 |
+
"execution_count": null,
|
6 |
+
"metadata": {},
|
7 |
+
"outputs": [],
|
8 |
+
"source": [
|
9 |
+
"%pip install sentence-transformers==2.0.0"
|
10 |
+
]
|
11 |
+
},
|
12 |
+
{
|
13 |
+
"attachments": {},
|
14 |
+
"cell_type": "markdown",
|
15 |
+
"metadata": {},
|
16 |
+
"source": [
|
17 |
+
"https://www.kaggle.com/datasets/dataranch/upwork-1"
|
18 |
+
]
|
19 |
+
},
|
20 |
+
{
|
21 |
+
"cell_type": "markdown",
|
22 |
+
"metadata": {},
|
23 |
+
"source": [
|
24 |
+
"1. Load dataset with pandas"
|
25 |
+
]
|
26 |
+
},
|
27 |
+
{
|
28 |
+
"cell_type": "code",
|
29 |
+
"execution_count": 135,
|
30 |
+
"metadata": {},
|
31 |
+
"outputs": [],
|
32 |
+
"source": [
|
33 |
+
"import pandas as pd\n",
|
34 |
+
"\n",
|
35 |
+
"df = pd.read_csv('freelancers.csv')\n",
|
36 |
+
"df = df[['shortName', 'title', 'description', 'location', 'hourlyRate', 'avgFeedbackScore', 'skills']]\n",
|
37 |
+
"df = df.dropna(subset='skills')\n",
|
38 |
+
"df = df.dropna(subset='title')\n",
|
39 |
+
"df['location'] = df['location'].apply(lambda x : eval(x)['state'])\n",
|
40 |
+
"df['skills'] = df['skills'].apply(lambda x : [x['skill']['name'] for x in eval(x)])\n",
|
41 |
+
"df['hourlyRate'] = df['hourlyRate'].apply(lambda x : eval(x)['amount'])"
|
42 |
+
]
|
43 |
+
},
|
44 |
+
{
|
45 |
+
"cell_type": "code",
|
46 |
+
"execution_count": null,
|
47 |
+
"metadata": {},
|
48 |
+
"outputs": [],
|
49 |
+
"source": [
|
50 |
+
"def keyword_extractor(total_keywords, str1):\n",
|
51 |
+
" word_list = list()\n",
|
52 |
+
" for keyword in total_keywords:\n",
|
53 |
+
" for word in str1.split(' '):\n",
|
54 |
+
" if word == keyword:\n",
|
55 |
+
" word_list.append(word)\n",
|
56 |
+
" return word_list\n",
|
57 |
+
"total_keywords = df.explode('skills')['skills'].unique().tolist()\n",
|
58 |
+
"\n",
|
59 |
+
"skill_keywords = keyword_extractor(total_keywords, 'I want to hire a wordpress')\n",
|
60 |
+
"df[df['skills'].apply(lambda x: all(val in x for val in skill_keywords))]"
|
61 |
+
]
|
62 |
+
},
|
63 |
+
{
|
64 |
+
"cell_type": "markdown",
|
65 |
+
"metadata": {},
|
66 |
+
"source": [
|
67 |
+
"2. Encode 100 samples into vectors (1 column with product text, 1 column with vectors)"
|
68 |
+
]
|
69 |
+
},
|
70 |
+
{
|
71 |
+
"cell_type": "code",
|
72 |
+
"execution_count": 136,
|
73 |
+
"metadata": {},
|
74 |
+
"outputs": [
|
75 |
+
{
|
76 |
+
"name": "stderr",
|
77 |
+
"output_type": "stream",
|
78 |
+
"text": [
|
79 |
+
"100%|██████████| 5815/5815 [12:10<00:00, 7.96it/s]\n"
|
80 |
+
]
|
81 |
+
},
|
82 |
+
{
|
83 |
+
"data": {
|
84 |
+
"text/html": [
|
85 |
+
"<div>\n",
|
86 |
+
"<style scoped>\n",
|
87 |
+
" .dataframe tbody tr th:only-of-type {\n",
|
88 |
+
" vertical-align: middle;\n",
|
89 |
+
" }\n",
|
90 |
+
"\n",
|
91 |
+
" .dataframe tbody tr th {\n",
|
92 |
+
" vertical-align: top;\n",
|
93 |
+
" }\n",
|
94 |
+
"\n",
|
95 |
+
" .dataframe thead th {\n",
|
96 |
+
" text-align: right;\n",
|
97 |
+
" }\n",
|
98 |
+
"</style>\n",
|
99 |
+
"<table border=\"1\" class=\"dataframe\">\n",
|
100 |
+
" <thead>\n",
|
101 |
+
" <tr style=\"text-align: right;\">\n",
|
102 |
+
" <th></th>\n",
|
103 |
+
" <th>shortName</th>\n",
|
104 |
+
" <th>title</th>\n",
|
105 |
+
" <th>description</th>\n",
|
106 |
+
" <th>location</th>\n",
|
107 |
+
" <th>hourlyRate</th>\n",
|
108 |
+
" <th>avgFeedbackScore</th>\n",
|
109 |
+
" <th>skills</th>\n",
|
110 |
+
" <th>text_vector_</th>\n",
|
111 |
+
" </tr>\n",
|
112 |
+
" </thead>\n",
|
113 |
+
" <tbody>\n",
|
114 |
+
" <tr>\n",
|
115 |
+
" <th>0</th>\n",
|
116 |
+
" <td>Jason V.</td>\n",
|
117 |
+
" <td>Expert WordPress Developer</td>\n",
|
118 |
+
" <td>Hello! Welcome to my profile!\\n\\nMy name is Ja...</td>\n",
|
119 |
+
" <td>IL</td>\n",
|
120 |
+
" <td>60.00</td>\n",
|
121 |
+
" <td>4.925208</td>\n",
|
122 |
+
" <td>[wordpress, seo, wp-ecommerce, woocommerce, bo...</td>\n",
|
123 |
+
" <td>[0.078628771007061, 0.024731114506721497, -0.0...</td>\n",
|
124 |
+
" </tr>\n",
|
125 |
+
" <tr>\n",
|
126 |
+
" <th>1</th>\n",
|
127 |
+
" <td>Miranda S.</td>\n",
|
128 |
+
" <td>Social Media Manager</td>\n",
|
129 |
+
" <td>I am a Social Media Manager who specializes in...</td>\n",
|
130 |
+
" <td>NY</td>\n",
|
131 |
+
" <td>20.00</td>\n",
|
132 |
+
" <td>4.675676</td>\n",
|
133 |
+
" <td>[social-media-content-creation, video-editing,...</td>\n",
|
134 |
+
" <td>[0.07423530519008636, -0.022386642172932625, -...</td>\n",
|
135 |
+
" </tr>\n",
|
136 |
+
" <tr>\n",
|
137 |
+
" <th>2</th>\n",
|
138 |
+
" <td>Gagan S J.</td>\n",
|
139 |
+
" <td>Solution Architect</td>\n",
|
140 |
+
" <td>More than 25 years in IT with 20 years in US h...</td>\n",
|
141 |
+
" <td>NJ</td>\n",
|
142 |
+
" <td>65.00</td>\n",
|
143 |
+
" <td>0.000000</td>\n",
|
144 |
+
" <td>[oracle-java-ee, spring-framework, hibernate, ...</td>\n",
|
145 |
+
" <td>[0.04637446999549866, 0.03554175794124603, -0....</td>\n",
|
146 |
+
" </tr>\n",
|
147 |
+
" <tr>\n",
|
148 |
+
" <th>3</th>\n",
|
149 |
+
" <td>Roxana L.</td>\n",
|
150 |
+
" <td>Procurement, Logistics and Supply Chain profes...</td>\n",
|
151 |
+
" <td>I work FAST - I am COST EFFICIENT - I deliver ...</td>\n",
|
152 |
+
" <td>NC</td>\n",
|
153 |
+
" <td>70.00</td>\n",
|
154 |
+
" <td>4.916684</td>\n",
|
155 |
+
" <td>[procurement-function, pharmaceutical-industry...</td>\n",
|
156 |
+
" <td>[0.026502298191189766, -0.02052873745560646, -...</td>\n",
|
157 |
+
" </tr>\n",
|
158 |
+
" <tr>\n",
|
159 |
+
" <th>4</th>\n",
|
160 |
+
" <td>John M.</td>\n",
|
161 |
+
" <td>3d generalist, post production artist and Vide...</td>\n",
|
162 |
+
" <td>I am a 3D artist, animator, and designer with ...</td>\n",
|
163 |
+
" <td>OK</td>\n",
|
164 |
+
" <td>50.00</td>\n",
|
165 |
+
" <td>5.000000</td>\n",
|
166 |
+
" <td>[animation, motion-graphics, video-editing, vi...</td>\n",
|
167 |
+
" <td>[0.05356863886117935, 0.032190944999456406, -0...</td>\n",
|
168 |
+
" </tr>\n",
|
169 |
+
" <tr>\n",
|
170 |
+
" <th>...</th>\n",
|
171 |
+
" <td>...</td>\n",
|
172 |
+
" <td>...</td>\n",
|
173 |
+
" <td>...</td>\n",
|
174 |
+
" <td>...</td>\n",
|
175 |
+
" <td>...</td>\n",
|
176 |
+
" <td>...</td>\n",
|
177 |
+
" <td>...</td>\n",
|
178 |
+
" <td>...</td>\n",
|
179 |
+
" </tr>\n",
|
180 |
+
" <tr>\n",
|
181 |
+
" <th>5818</th>\n",
|
182 |
+
" <td>Megan D.</td>\n",
|
183 |
+
" <td>Web Research/Content Curation/Data Entry/Socia...</td>\n",
|
184 |
+
" <td>Extremely driven, yet easy-going gal who refus...</td>\n",
|
185 |
+
" <td>CA</td>\n",
|
186 |
+
" <td>22.37</td>\n",
|
187 |
+
" <td>4.574029</td>\n",
|
188 |
+
" <td>[data-entry, internet-research, virtual-assist...</td>\n",
|
189 |
+
" <td>[0.07812528312206268, -0.018792806193232536, -...</td>\n",
|
190 |
+
" </tr>\n",
|
191 |
+
" <tr>\n",
|
192 |
+
" <th>5819</th>\n",
|
193 |
+
" <td>Austin V.</td>\n",
|
194 |
+
" <td>Product Manager</td>\n",
|
195 |
+
" <td>Experienced in building and growing digital pl...</td>\n",
|
196 |
+
" <td>AZ</td>\n",
|
197 |
+
" <td>100.00</td>\n",
|
198 |
+
" <td>0.000000</td>\n",
|
199 |
+
" <td>[atlassian-jira, atlassian-confluence, project...</td>\n",
|
200 |
+
" <td>[0.056266412138938904, -0.007661229465156794, ...</td>\n",
|
201 |
+
" </tr>\n",
|
202 |
+
" <tr>\n",
|
203 |
+
" <th>5820</th>\n",
|
204 |
+
" <td>Eric M.</td>\n",
|
205 |
+
" <td>Web Developer/Designer And Graphic Designer</td>\n",
|
206 |
+
" <td>4 YEARS experience in Wordpress / Shopify / D...</td>\n",
|
207 |
+
" <td>OH</td>\n",
|
208 |
+
" <td>25.00</td>\n",
|
209 |
+
" <td>4.451507</td>\n",
|
210 |
+
" <td>[html, css, wordpress, shopify, joomla, drupal...</td>\n",
|
211 |
+
" <td>[0.048749279230833054, -0.013894445262849331, ...</td>\n",
|
212 |
+
" </tr>\n",
|
213 |
+
" <tr>\n",
|
214 |
+
" <th>5821</th>\n",
|
215 |
+
" <td>Kristina A.</td>\n",
|
216 |
+
" <td>Voice Over Talent, video editing, video produc...</td>\n",
|
217 |
+
" <td>A musician from birth. I studied music educati...</td>\n",
|
218 |
+
" <td>VA</td>\n",
|
219 |
+
" <td>60.00</td>\n",
|
220 |
+
" <td>5.000000</td>\n",
|
221 |
+
" <td>[articulate]</td>\n",
|
222 |
+
" <td>[0.03207482025027275, -0.027680398896336555, -...</td>\n",
|
223 |
+
" </tr>\n",
|
224 |
+
" <tr>\n",
|
225 |
+
" <th>5822</th>\n",
|
226 |
+
" <td>Ashley C.</td>\n",
|
227 |
+
" <td>Print and Digital Layout Designer</td>\n",
|
228 |
+
" <td>Hello, I'm Ashley, a professional designer and...</td>\n",
|
229 |
+
" <td>OR</td>\n",
|
230 |
+
" <td>25.00</td>\n",
|
231 |
+
" <td>4.951186</td>\n",
|
232 |
+
" <td>[print-layout-design, brochure-design, flyer-d...</td>\n",
|
233 |
+
" <td>[0.04141489043831825, -0.04500063508749008, -0...</td>\n",
|
234 |
+
" </tr>\n",
|
235 |
+
" </tbody>\n",
|
236 |
+
"</table>\n",
|
237 |
+
"<p>5815 rows × 8 columns</p>\n",
|
238 |
+
"</div>"
|
239 |
+
],
|
240 |
+
"text/plain": [
|
241 |
+
" shortName title \\\n",
|
242 |
+
"0 Jason V. Expert WordPress Developer \n",
|
243 |
+
"1 Miranda S. Social Media Manager \n",
|
244 |
+
"2 Gagan S J. Solution Architect \n",
|
245 |
+
"3 Roxana L. Procurement, Logistics and Supply Chain profes... \n",
|
246 |
+
"4 John M. 3d generalist, post production artist and Vide... \n",
|
247 |
+
"... ... ... \n",
|
248 |
+
"5818 Megan D. Web Research/Content Curation/Data Entry/Socia... \n",
|
249 |
+
"5819 Austin V. Product Manager \n",
|
250 |
+
"5820 Eric M. Web Developer/Designer And Graphic Designer \n",
|
251 |
+
"5821 Kristina A. Voice Over Talent, video editing, video produc... \n",
|
252 |
+
"5822 Ashley C. Print and Digital Layout Designer \n",
|
253 |
+
"\n",
|
254 |
+
" description location hourlyRate \\\n",
|
255 |
+
"0 Hello! Welcome to my profile!\\n\\nMy name is Ja... IL 60.00 \n",
|
256 |
+
"1 I am a Social Media Manager who specializes in... NY 20.00 \n",
|
257 |
+
"2 More than 25 years in IT with 20 years in US h... NJ 65.00 \n",
|
258 |
+
"3 I work FAST - I am COST EFFICIENT - I deliver ... NC 70.00 \n",
|
259 |
+
"4 I am a 3D artist, animator, and designer with ... OK 50.00 \n",
|
260 |
+
"... ... ... ... \n",
|
261 |
+
"5818 Extremely driven, yet easy-going gal who refus... CA 22.37 \n",
|
262 |
+
"5819 Experienced in building and growing digital pl... AZ 100.00 \n",
|
263 |
+
"5820 4 YEARS experience in Wordpress / Shopify / D... OH 25.00 \n",
|
264 |
+
"5821 A musician from birth. I studied music educati... VA 60.00 \n",
|
265 |
+
"5822 Hello, I'm Ashley, a professional designer and... OR 25.00 \n",
|
266 |
+
"\n",
|
267 |
+
" avgFeedbackScore skills \\\n",
|
268 |
+
"0 4.925208 [wordpress, seo, wp-ecommerce, woocommerce, bo... \n",
|
269 |
+
"1 4.675676 [social-media-content-creation, video-editing,... \n",
|
270 |
+
"2 0.000000 [oracle-java-ee, spring-framework, hibernate, ... \n",
|
271 |
+
"3 4.916684 [procurement-function, pharmaceutical-industry... \n",
|
272 |
+
"4 5.000000 [animation, motion-graphics, video-editing, vi... \n",
|
273 |
+
"... ... ... \n",
|
274 |
+
"5818 4.574029 [data-entry, internet-research, virtual-assist... \n",
|
275 |
+
"5819 0.000000 [atlassian-jira, atlassian-confluence, project... \n",
|
276 |
+
"5820 4.451507 [html, css, wordpress, shopify, joomla, drupal... \n",
|
277 |
+
"5821 5.000000 [articulate] \n",
|
278 |
+
"5822 4.951186 [print-layout-design, brochure-design, flyer-d... \n",
|
279 |
+
"\n",
|
280 |
+
" text_vector_ \n",
|
281 |
+
"0 [0.078628771007061, 0.024731114506721497, -0.0... \n",
|
282 |
+
"1 [0.07423530519008636, -0.022386642172932625, -... \n",
|
283 |
+
"2 [0.04637446999549866, 0.03554175794124603, -0.... \n",
|
284 |
+
"3 [0.026502298191189766, -0.02052873745560646, -... \n",
|
285 |
+
"4 [0.05356863886117935, 0.032190944999456406, -0... \n",
|
286 |
+
"... ... \n",
|
287 |
+
"5818 [0.07812528312206268, -0.018792806193232536, -... \n",
|
288 |
+
"5819 [0.056266412138938904, -0.007661229465156794, ... \n",
|
289 |
+
"5820 [0.048749279230833054, -0.013894445262849331, ... \n",
|
290 |
+
"5821 [0.03207482025027275, -0.027680398896336555, -... \n",
|
291 |
+
"5822 [0.04141489043831825, -0.04500063508749008, -0... \n",
|
292 |
+
"\n",
|
293 |
+
"[5815 rows x 8 columns]"
|
294 |
+
]
|
295 |
+
},
|
296 |
+
"execution_count": 136,
|
297 |
+
"metadata": {},
|
298 |
+
"output_type": "execute_result"
|
299 |
+
}
|
300 |
+
],
|
301 |
+
"source": [
|
302 |
+
"import pandas as pd\n",
|
303 |
+
"from tqdm import tqdm\n",
|
304 |
+
"from sentence_transformers import SentenceTransformer\n",
|
305 |
+
"tqdm.pandas()\n",
|
306 |
+
"\n",
|
307 |
+
"model = SentenceTransformer('all-mpnet-base-v2') #all-MiniLM-L6-v2 #all-mpnet-base-v2\n",
|
308 |
+
"\n",
|
309 |
+
"#encode df version: for small dataset only\n",
|
310 |
+
"df['text_vector_'] = df['title'].progress_apply(lambda x : model.encode(x).tolist())\n",
|
311 |
+
"df"
|
312 |
+
]
|
313 |
+
},
|
314 |
+
{
|
315 |
+
"cell_type": "code",
|
316 |
+
"execution_count": 133,
|
317 |
+
"metadata": {},
|
318 |
+
"outputs": [],
|
319 |
+
"source": [
|
320 |
+
"import random\n",
|
321 |
+
"\n",
|
322 |
+
"df = df.drop_duplicates(subset=['shortName', 'location', 'title', 'hourlyRate', 'avgFeedbackScore', 'description'])\n",
|
323 |
+
"df = df.reset_index(drop=True)\n",
|
324 |
+
"df['location'] = df['location'].apply(lambda x : random.choice(['New York', 'Chicago', 'Washington']))\n",
|
325 |
+
"df.to_parquet('df_encoded.parquet', index=None)"
|
326 |
+
]
|
327 |
+
}
|
328 |
+
],
|
329 |
+
"metadata": {
|
330 |
+
"kernelspec": {
|
331 |
+
"display_name": "Python 3.9.0 64-bit",
|
332 |
+
"language": "python",
|
333 |
+
"name": "python3"
|
334 |
+
},
|
335 |
+
"language_info": {
|
336 |
+
"codemirror_mode": {
|
337 |
+
"name": "ipython",
|
338 |
+
"version": 3
|
339 |
+
},
|
340 |
+
"file_extension": ".py",
|
341 |
+
"mimetype": "text/x-python",
|
342 |
+
"name": "python",
|
343 |
+
"nbconvert_exporter": "python",
|
344 |
+
"pygments_lexer": "ipython3",
|
345 |
+
"version": "3.9.13"
|
346 |
+
},
|
347 |
+
"orig_nbformat": 4,
|
348 |
+
"vscode": {
|
349 |
+
"interpreter": {
|
350 |
+
"hash": "fdf377d643bc1cb065454f0ad2ceac75d834452ecf289e7ba92c6b3f59a7cee1"
|
351 |
+
}
|
352 |
+
}
|
353 |
+
},
|
354 |
+
"nbformat": 4,
|
355 |
+
"nbformat_minor": 2
|
356 |
+
}
|