1.py

# PRELIMINARY OPERATIONS
data = pd.read_excel("./data_original.xlsx")

data = data.drop(columns=["fix_port","source", "age_youngest", "gender", "injury_desc", "report", "notes", "mechanical", "op_error", "employee"])
data = data.dropna()

data = data[data["bus_type"]=="Amusement park"]
data = data[data["industry_sector"]=="amusement ride"]
data = data[data["manufacturer"] != "In-house"]

data = data.drop(columns=["bus_type", "industry_sector"])

data.reset_index(inplace=True)
data.drop(columns="index", inplace=True)