Skip to content

Commit 122772a

Browse files
author
bitoollearner
committed
LeetCode Pyspark Solution
1 parent 308bb5f commit 122772a

11 files changed

+1221
-82
lines changed

‎Solved/2020. Number of Accounts That Did Not Stream (Medium)-(Solved).ipynb

Lines changed: 159 additions & 13 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,10 @@
44
"cell_type": "markdown",
55
"metadata": {
66
"application/vnd.databricks.v1+cell": {
7-
"cellMetadata": {},
7+
"cellMetadata": {
8+
"byteLimit": 2048000,
9+
"rowLimit": 10000
10+
},
811
"inputWidgets": {},
912
"nuid": "83d38679-f742-44bb-952b-b541e178ddc8",
1013
"showTitle": false,
@@ -21,7 +24,10 @@
2124
"execution_count": 0,
2225
"metadata": {
2326
"application/vnd.databricks.v1+cell": {
24-
"cellMetadata": {},
27+
"cellMetadata": {
28+
"byteLimit": 2048000,
29+
"rowLimit": 10000
30+
},
2531
"inputWidgets": {},
2632
"nuid": "1daf4387-60f2-4c66-b119-d9ce2cd54aa2",
2733
"showTitle": false,
@@ -40,7 +46,10 @@
4046
"cell_type": "markdown",
4147
"metadata": {
4248
"application/vnd.databricks.v1+cell": {
43-
"cellMetadata": {},
49+
"cellMetadata": {
50+
"byteLimit": 2048000,
51+
"rowLimit": 10000
52+
},
4453
"inputWidgets": {},
4554
"nuid": "83ab9b20-fd4f-4392-ae0c-ff0580c3c7c1",
4655
"showTitle": false,
@@ -86,11 +95,11 @@
8695
"**Subscriptions table:**\n",
8796
"| account_id | start_date | end_date |\n",
8897
"|------------|------------|------------|\n",
89-
"| 9 | 2020-02-18 | 2021-10-30 |\n",
98+
"| 9 | 2021-02-18 | 2021-10-30 |\n",
9099
"| 3 | 2021-09-21 | 2021-11-13 |\n",
91100
"| 11 | 2020-02-28 | 2020-08-18 |\n",
92101
"| 13 | 2021-04-20 | 2021-09-22 |\n",
93-
"| 4 | 2020-10-26 | 2021-05-08 |\n",
102+
"| 4 | 2021-01-26 | 2021-05-08 |\n",
94103
"| 5 | 2020-09-11 | 2021-01-17 |\n",
95104
"\n",
96105
"**Streams table:**\n",
@@ -117,23 +126,35 @@
117126
"execution_count": 0,
118127
"metadata": {
119128
"application/vnd.databricks.v1+cell": {
120-
"cellMetadata": {},
129+
"cellMetadata": {
130+
"byteLimit": 2048000,
131+
"rowLimit": 10000
132+
},
121133
"inputWidgets": {},
122134
"nuid": "e39af07f-85df-4e21-8a5f-c65604a811f3",
123135
"showTitle": false,
124136
"tableResultSettingsMap": {},
125137
"title": ""
126138
}
127139
},
128-
"outputs": [],
140+
"outputs": [
141+
{
142+
"output_type": "stream",
143+
"name": "stdout",
144+
"output_type": "stream",
145+
"text": [
146+
"+----------+----------+----------+\n|account_id|start_date| end_date|\n+----------+----------+----------+\n| 9|2021-02-18|2021-10-30|\n| 3|2021-09-21|2021-11-13|\n| 11|2020-02-28|2020-08-18|\n| 13|2021-04-20|2021-09-22|\n| 4|2021-01-26|2021-05-08|\n| 5|2020-09-11|2021-01-17|\n+----------+----------+----------+\n\n+----------+----------+-----------+\n|session_id|account_id|stream_date|\n+----------+----------+-----------+\n| 14| 9| 2020-05-16|\n| 16| 3| 2021-10-27|\n| 18| 11| 2020-04-29|\n| 17| 13| 2021-08-08|\n| 19| 4| 2020-12-31|\n| 13| 5| 2021-01-05|\n+----------+----------+-----------+\n\n"
147+
]
148+
}
149+
],
129150
"source": [
130151
"subscriptions_data_2020 = [\n",
131-
" (9, \"2020-02-18\", \"2021-10-30\"),\n",
152+
" (9, \"2021-02-18\", \"2021-10-30\"),\n",
132153
" (3, \"2021-09-21\", \"2021-11-13\"),\n",
133154
" (11, \"2020-02-28\", \"2020-08-18\"),\n",
134155
" (13, \"2021-04-20\", \"2021-09-22\"),\n",
135-
" (4, \"2020-10-26\", \"2021-05-08\"),\n",
136-
" (5, \"2020-09-11\", \"2021-01-17\"),\n",
156+
" (4, \"2021-01-26\", \"2021-05-08\"),\n",
157+
" (5, \"2020-09-11\", \"2021-01-17\")\n",
137158
"]\n",
138159
"\n",
139160
"subscriptions_columns_2020 = [\"account_id\", \"start_date\", \"end_date\"]\n",
@@ -146,22 +167,147 @@
146167
" (18, 11, \"2020-04-29\"),\n",
147168
" (17, 13, \"2021-08-08\"),\n",
148169
" (19, 4, \"2020-12-31\"),\n",
149-
" (13, 5, \"2021-01-05\"),\n",
170+
" (13, 5, \"2021-01-05\")\n",
150171
"]\n",
151172
"\n",
152173
"streams_columns_2020 = [\"session_id\", \"account_id\", \"stream_date\"]\n",
153174
"streams_df_2020 = spark.createDataFrame(streams_data_2020, streams_columns_2020)\n",
154175
"streams_df_2020.show()"
155176
]
177+
},
178+
{
179+
"cell_type": "code",
180+
"execution_count": 0,
181+
"metadata": {
182+
"application/vnd.databricks.v1+cell": {
183+
"cellMetadata": {
184+
"byteLimit": 2048000,
185+
"rowLimit": 10000
186+
},
187+
"inputWidgets": {},
188+
"nuid": "0901c1c2-69b2-4da0-b035-e904e9da7390",
189+
"showTitle": false,
190+
"tableResultSettingsMap": {},
191+
"title": ""
192+
}
193+
},
194+
"outputs": [],
195+
"source": [
196+
"subscriptions_df_2020 = subscriptions_df_2020\\\n",
197+
" .withColumn(\"start_date\", col(\"start_date\").cast(\"date\")) \\\n",
198+
" .withColumn(\"end_date\", col(\"end_date\").cast(\"date\"))"
199+
]
200+
},
201+
{
202+
"cell_type": "code",
203+
"execution_count": 0,
204+
"metadata": {
205+
"application/vnd.databricks.v1+cell": {
206+
"cellMetadata": {
207+
"byteLimit": 2048000,
208+
"rowLimit": 10000
209+
},
210+
"inputWidgets": {},
211+
"nuid": "903bc630-1067-4606-a2b3-05b63ed59c29",
212+
"showTitle": false,
213+
"tableResultSettingsMap": {},
214+
"title": ""
215+
}
216+
},
217+
"outputs": [],
218+
"source": [
219+
"streams_df_2020 = streams_df_2020\\\n",
220+
" .withColumn(\"stream_date\", col(\"stream_date\").cast(\"date\"))"
221+
]
222+
},
223+
{
224+
"cell_type": "code",
225+
"execution_count": 0,
226+
"metadata": {
227+
"application/vnd.databricks.v1+cell": {
228+
"cellMetadata": {
229+
"byteLimit": 2048000,
230+
"rowLimit": 10000
231+
},
232+
"inputWidgets": {},
233+
"nuid": "4b4707d2-78f1-464e-8d32-84556314f070",
234+
"showTitle": false,
235+
"tableResultSettingsMap": {},
236+
"title": ""
237+
}
238+
},
239+
"outputs": [],
240+
"source": [
241+
"subs_2021 = subscriptions_df_2020.filter(year(\"start_date\") == 2021)"
242+
]
243+
},
244+
{
245+
"cell_type": "code",
246+
"execution_count": 0,
247+
"metadata": {
248+
"application/vnd.databricks.v1+cell": {
249+
"cellMetadata": {
250+
"byteLimit": 2048000,
251+
"rowLimit": 10000
252+
},
253+
"inputWidgets": {},
254+
"nuid": "3cc33f83-171a-4bc0-8f34-b5ede8d4f5ef",
255+
"showTitle": false,
256+
"tableResultSettingsMap": {},
257+
"title": ""
258+
}
259+
},
260+
"outputs": [],
261+
"source": [
262+
"streams_2021 = streams_df_2020.filter(year(\"stream_date\") == 2021)"
263+
]
264+
},
265+
{
266+
"cell_type": "code",
267+
"execution_count": 0,
268+
"metadata": {
269+
"application/vnd.databricks.v1+cell": {
270+
"cellMetadata": {
271+
"byteLimit": 2048000,
272+
"rowLimit": 10000
273+
},
274+
"inputWidgets": {},
275+
"nuid": "528f1f88-1726-41d2-a401-51727e965d43",
276+
"showTitle": false,
277+
"tableResultSettingsMap": {},
278+
"title": ""
279+
}
280+
},
281+
"outputs": [
282+
{
283+
"output_type": "stream",
284+
"name": "stdout",
285+
"output_type": "stream",
286+
"text": [
287+
"+--------------+\n|accounts_count|\n+--------------+\n| 2|\n+--------------+\n\n"
288+
]
289+
}
290+
],
291+
"source": [
292+
"subs_2021.join(streams_2021, \"account_id\", \"left\")\\\n",
293+
" .filter(streams_2021[\"session_id\"].isNull())\\\n",
294+
" .select(countDistinct(\"account_id\").alias(\"accounts_count\")).show()"
295+
]
156296
}
157297
],
158298
"metadata": {
159299
"application/vnd.databricks.v1+notebook": {
160-
"computePreferences": null,
300+
"computePreferences": {
301+
"hardware": {
302+
"accelerator": null,
303+
"gpuPoolId": null,
304+
"memory": null
305+
}
306+
},
161307
"dashboards": [],
162308
"environmentMetadata": {
163309
"base_environment": "",
164-
"environment_version": "1"
310+
"environment_version": "2"
165311
},
166312
"inputWidgetPreferences": null,
167313
"language": "python",

‎Solved/2026. Low-Quality Problems (Easy)-(Solved).ipynb

Lines changed: 67 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -4,7 +4,10 @@
44
"cell_type": "markdown",
55
"metadata": {
66
"application/vnd.databricks.v1+cell": {
7-
"cellMetadata": {},
7+
"cellMetadata": {
8+
"byteLimit": 2048000,
9+
"rowLimit": 10000
10+
},
811
"inputWidgets": {},
912
"nuid": "83d38679-f742-44bb-952b-b541e178ddc8",
1013
"showTitle": false,
@@ -21,7 +24,10 @@
2124
"execution_count": 0,
2225
"metadata": {
2326
"application/vnd.databricks.v1+cell": {
24-
"cellMetadata": {},
27+
"cellMetadata": {
28+
"byteLimit": 2048000,
29+
"rowLimit": 10000
30+
},
2531
"inputWidgets": {},
2632
"nuid": "1daf4387-60f2-4c66-b119-d9ce2cd54aa2",
2733
"showTitle": false,
@@ -40,7 +46,10 @@
4046
"cell_type": "markdown",
4147
"metadata": {
4248
"application/vnd.databricks.v1+cell": {
43-
"cellMetadata": {},
49+
"cellMetadata": {
50+
"byteLimit": 2048000,
51+
"rowLimit": 10000
52+
},
4453
"inputWidgets": {},
4554
"nuid": "83ab9b20-fd4f-4392-ae0c-ff0580c3c7c1",
4655
"showTitle": false,
@@ -106,15 +115,27 @@
106115
"execution_count": 0,
107116
"metadata": {
108117
"application/vnd.databricks.v1+cell": {
109-
"cellMetadata": {},
118+
"cellMetadata": {
119+
"byteLimit": 2048000,
120+
"rowLimit": 10000
121+
},
110122
"inputWidgets": {},
111123
"nuid": "e39af07f-85df-4e21-8a5f-c65604a811f3",
112124
"showTitle": false,
113125
"tableResultSettingsMap": {},
114126
"title": ""
115127
}
116128
},
117-
"outputs": [],
129+
"outputs": [
130+
{
131+
"output_type": "stream",
132+
"name": "stdout",
133+
"output_type": "stream",
134+
"text": [
135+
"+----------+-----+--------+\n|problem_id|likes|dislikes|\n+----------+-----+--------+\n| 6| 1290| 425|\n| 11| 2677| 8659|\n| 1| 4446| 2760|\n| 7| 8569| 6086|\n| 13| 2050| 4164|\n| 10| 9002| 7446|\n+----------+-----+--------+\n\n"
136+
]
137+
}
138+
],
118139
"source": [
119140
"problems_data_2026 = [\n",
120141
" (6, 1290, 425),\n",
@@ -129,15 +150,54 @@
129150
"problems_df_2026 = spark.createDataFrame(problems_data_2026, problems_columns_2026)\n",
130151
"problems_df_2026.show()"
131152
]
153+
},
154+
{
155+
"cell_type": "code",
156+
"execution_count": 0,
157+
"metadata": {
158+
"application/vnd.databricks.v1+cell": {
159+
"cellMetadata": {
160+
"byteLimit": 2048000,
161+
"rowLimit": 10000
162+
},
163+
"inputWidgets": {},
164+
"nuid": "c1c4d9d1-4cf7-4cd5-8c7b-e096f1da08cd",
165+
"showTitle": false,
166+
"tableResultSettingsMap": {},
167+
"title": ""
168+
}
169+
},
170+
"outputs": [
171+
{
172+
"output_type": "stream",
173+
"name": "stdout",
174+
"output_type": "stream",
175+
"text": [
176+
"+----------+\n|problem_id|\n+----------+\n| 7|\n| 10|\n| 11|\n| 13|\n+----------+\n\n"
177+
]
178+
}
179+
],
180+
"source": [
181+
"problems_df_2026\\\n",
182+
" .withColumn( \"like_percentage\", col(\"likes\") / (col(\"likes\") + col(\"dislikes\")))\\\n",
183+
" .filter(col(\"like_percentage\") < 0.6)\\\n",
184+
" .select(\"problem_id\").orderBy(\"problem_id\").show()"
185+
]
132186
}
133187
],
134188
"metadata": {
135189
"application/vnd.databricks.v1+notebook": {
136-
"computePreferences": null,
190+
"computePreferences": {
191+
"hardware": {
192+
"accelerator": null,
193+
"gpuPoolId": null,
194+
"memory": null
195+
}
196+
},
137197
"dashboards": [],
138198
"environmentMetadata": {
139199
"base_environment": "",
140-
"environment_version": "1"
200+
"environment_version": "2"
141201
},
142202
"inputWidgetPreferences": null,
143203
"language": "python",

0 commit comments

Comments
 (0)