Google の ML を使用した予想 のレビュー

Kumar Raj Deepak · 約2年前にレビュー済み

Shrivastava Ayushi · 約2年前にレビュー済み

S SUMESHA · 約2年前にレビュー済み

Mense Akshata · 約2年前にレビュー済み

Kumari Anjali · 約2年前にレビュー済み

Katkade Siddhesh · 約2年前にレビュー済み

Marathe Sangeeta · 約2年前にレビュー済み

Gupta Meena · 約2年前にレビュー済み

nice

Kumar Hitesh · 約2年前にレビュー済み

Dokwal Ayushi · 約2年前にレビュー済み

Great

RAJ ALOK · 約2年前にレビュー済み

jaju ronit · 約2年前にレビュー済み

NICE

Nag Tiyasha · 約2年前にレビュー済み

Jain Abhishek · 約2年前にレビュー済み

Nath Amrit · 約2年前にレビュー済み

nice

Chavan Vaishnavi · 約2年前にレビュー済み

ok

Kumar Mohit · 約2年前にレビュー済み

Shukla Pratiksha · 約2年前にレビュー済み

Venkata Sai Santoshi Arji · 約2年前にレビュー済み

Choudhury Suvendu · 約2年前にレビュー済み

The SQL command in Task 12 is incorrect. The correct one would be: # create training dataset: # create a row for the winning team CREATE OR REPLACE TABLE `bracketology.training_new_features` AS WITH outcomes AS ( SELECT # features season, # 1994 "win" AS label, # our label win_seed AS seed, # ranking # this time without seed even win_school_ncaa AS school_ncaa, lose_seed AS opponent_seed, # ranking lose_school_ncaa AS opponent_school_ncaa FROM `bigquery-public-data.ncaa_basketball.mbb_historical_tournament_games` t WHERE season >= 2014 UNION ALL # create a separate row for the losing team SELECT # features season, # 1994 "loss" AS label, # our label lose_seed AS seed, # ranking lose_school_ncaa AS school_ncaa, win_seed AS opponent_seed, # ranking win_school_ncaa AS opponent_school_ncaa FROM `bigquery-public-data.ncaa_basketball.mbb_historical_tournament_games` t WHERE season >= 2014 UNION ALL # add in 2018 tournament game results not part of the public dataset: SELECT season, label, seed, school_ncaa, opponent_seed, opponent_school_ncaa FROM `data-to-insights.ncaa.2018_tournament_results` ) SELECT o.season, label, # our team seed, school_ncaa, # new pace metrics (basketball possession) team.pace_rank, team.poss_40min, team.pace_rating, # new efficiency metrics (scoring over time) team.efficiency_rank, team.pts_100poss, team.efficiency_rating, # opposing team opponent_seed, opponent_school_ncaa, # new pace metrics (basketball possession) opp.pace_rank AS opp_pace_rank, opp.poss_40min AS opp_poss_40min, opp.pace_rating AS opp_pace_rating, # new efficiency metrics (scoring over time) opp.efficiency_rank AS opp_efficiency_rank, opp.pts_100poss AS opp_pts_100poss, opp.efficiency_rating AS opp_efficiency_rating, # a little feature engineering (take the difference in stats) # new pace metrics (basketball possession) opp.pace_rank - team.pace_rank AS pace_rank_diff, opp.poss_40min - team.poss_40min AS pace_stat_diff, opp.pace_rating - team.pace_rating AS pace_rating_diff, # new efficiency metrics (scoring over time) opp.efficiency_rank - team.efficiency_rank AS eff_rank_diff, opp.pts_100poss - team.pts_100poss AS eff_stat_diff, opp.efficiency_rating - team.efficiency_rating AS eff_rating_diff FROM outcomes AS o LEFT JOIN `data-to-insights.ncaa.feature_engineering` AS team ON o.school_ncaa = team.team AND o.season = team.season LEFT JOIN `data-to-insights.ncaa.feature_engineering` AS opp ON o.opponent_school_ncaa = opp.team AND o.season = opp.season The SQL command in Task 15 is also incorrect. The correct one is: CREATE OR REPLACE MODEL `bracketology.ncaa_model_updated` OPTIONS ( model_type='logistic_reg') AS SELECT # this time, don't train the model on school name or seed season, label, # our pace poss_40min, pace_rank, pace_rating, # opponent pace opp_poss_40min, opp_pace_rank, opp_pace_rating, # difference in pace pace_rank_diff, pace_stat_diff, pace_rating_diff, # our efficiency pts_100poss, efficiency_rank, efficiency_rating, # opponent efficiency opp_pts_100poss, opp_efficiency_rank, opp_efficiency_rating, # difference in efficiency eff_rank_diff, eff_stat_diff, eff_rating_diff FROM `bracketology.training_new_features` # here we'll train on 2014 - 2017 and predict on 2018 WHERE season BETWEEN 2014 AND 2017 # between in SQL is inclusive of end points

Irfanuddin Chowdhury Mohammad · 約2年前にレビュー済み

MISHRA ARUNENDRA · 約2年前にレビュー済み

Bachu Dhyaneswar · 約2年前にレビュー済み

Ali Kaif · 約2年前にレビュー済み

GOOGLEUSER Abhishek · 約2年前にレビュー済み