PROBLEM
You will be given a training dataset which consists of bus data for 4 types of existing bus trips, namely square, circle, star, real_world.
Click here to download dataset (Refer to dataset_README.md)
Dataset for each of the above types of trips, is present in a csv file consisting of several attributes. Naming format is
<type>_<train/test>.csv, which is self explanatory.
What you need to submit is mentioned in the next slide.
A SUBMISSION
For each of the 5 test csv files namely, square_test, circle_test, star_test, real_1_test, real_2_test, follow steps 2-6. Steps 2-6 are illustrated for the square type in particular.
For the provided square_test.csv containing 50 data points, your task is to predict the next 25 bus locations, based on the 50 data points. Put these 25 predictions in a csv file named <team_name>_square.csv.
Now use the combined data (50 provided by us initially + 25 data points generated in step 1). Take the last 50 rows of this new dataset and predict the next 25 locations based on these. Append these 25 predictions to <team_name>_square.csv.
Now you will have totally (50 [initial] + 25 [from step 2] + 25 [from step 3] = 100) data points.
Now again use this data consisting of 100 data points, take the last 50 rows from this and predict the next 25 locations. Append these 25 predictions to <team_name>_square.csv.
Repeat steps 2-5 in a cyclic manner until you generate 250 data points (excluding the initial 50 we provided you). (There must be 250 rows in <team_name>_square.csv.)
Contact Us
Agrim Jain
+91 9528876421
210010003@iitdh.ac.in
Shashank P
+91 9483303320
200010048@iitdh.ac.in
