Simple Baseline solution for ML Hikeathon

Simple_baseline_solution.py

'''
# Baseline script by Supreet Manyam (Ziron)

'''

import pandas as pd
import numpy as np
import gc

train = pd.read_csv("data/train.csv", 
                    usecols=["node1_id", "node2_id", "is_chat"], 
                    dtype={"node1_id": np.int64, "node2_id":np.int64, "is_chat": np.int8})
test = pd.read_csv("data/test.csv", 
                    usecols=["id", "node1_id", "node2_id"], 
                    dtype={"id": np.int64, "node1_id": np.int64, "node2_id":np.int64})

node1 = train.groupby("node1_id", as_index=False)["is_chat"].mean()
node2 = train.groupby("node2_id", as_index=False)["is_chat"].mean()

del train
gc.collect()

test = (test.merge(node1, on="node1_id", how="left")
            .merge(node2, on="node2_id", how="left", suffixes=('_1', '_2')))

test["is_chat"] = test["is_chat_1"].fillna(0)*0.5 + test["is_chat_2"].fillna(0)*0.5
test[["id", "is_chat"]].to_csv("submit.csv", index=False)