add py script

format.py

@@ -0,0 +1,40 @@
+#it is used to clean and create the database
+import sqlite3
+import pandas
+conn = sqlite3.connect("data.sqlite3")
+cur = conn.cursor()
+df = pandas.read_csv("offencebymonth.csv")
+#drop empty value
+new_df = df.dropna()
+#get rid of spaces
+new_df.replace(' ', '_', regex=True)
+LGA = df['LGA'].unique()
+subdiv = df['Statistical Division or Subdivision'].unique()
+cat = df["Offence category"].unique()
+#database split to 3 tables, main statistics, subdiv index and statistics4
+#Enale foreign key support
+print(cat)
+
+cur.execute("PRAGMA foreign_keys = ON;")
+
+categories = """CREATE TABLE IF NOT EXISTS categories(id INTEGER AUTO INCREMENT PRIMARY KEY, name TEXT)""" 
+subcat = """CREATE TABLE IF NOT EXISTS subcat(id INTEGER AUTO INCREMENT PRIMARY KEY, name TEXT, cat_id INTEGER, FOREIGN KEY(cat_id) REFERENCES categories(id))"""
+location = """CREATE TABLE IF NOT EXISTS location(id INTEGER AUTO INCREMENT PRIMARY KEY, subdiv TEXT, LGA TEXT)"""
+stat = """CREATE TABLE IF NOT EXISTS statistics(id INTEGER AUTO INCREMENT PRIMARY KEY, category_id INTEGER, subcat_id INTEGER, location_id INTEGER, month TEXT NOT NULL, count INTEGER NOT NULL,FOREIGN KEY(category_id) REFERENCES categories(id), FOREIGN KEY(subcat_id) REFERENCES subcat(id), FOREIGN KEY(location_id) REFERENCES location(id))"""
+cur.execute(categories)
+cur.execute(subcat)
+cur.execute(location)
+cur.execute(stat)
+#insert statements
+categories_ins = """INSERT INTO categories VALUES(NULL, ?)"""
+subcat_ins = """INSERT INTO subcat VALUES(NULL, ?, ?)"""
+location_ins = """INSERT INTO location(NULL, ?, ?)"""
+stat_ins = """INSERT INTO statistics(NULL, ?,?,?,?,?)"""
+cat = cat.tolist()
+print(cat)
+for category in cat:
+    print(category)
+    cur.executemany(categories_ins, category)
+
+cur.execute("SELECT * FROM categories")
+print(cur.fetchall())