update

Author: HaoUNSW

Prompting/ecl.py

@@ -0,0 +1,73 @@
+import numpy as np
+import os
+import datetime
+
+
+def add_days(start_day, delta_days):
+    date_1 = datetime.datetime.strptime(start_day, '%B %d, %Y')
+    end_date = date_1 + datetime.timedelta(days=delta_days)
+    out = end_date.strftime('%B %d, %Y')
+
+    return out
+
+
+def generate_numerical(raw_folder, save_path, mode="test", obs_length=15):
+    raw_data = np.load(os.path.join(raw_folder, "ecl_raw_" + mode + ".npy"))
+    data_x = []
+    data_y = []
+    for i in range(len(raw_data)):
+        data = raw_data[i]
+        number_of_instance = len(data) - obs_length
+        for j in range(number_of_instance):
+            y = data[obs_length + j]
+            x = data[j: obs_length + j]
+            data_x.append(x)
+            data_y.append(y)
+
+    data_x = np.reshape(data_x, [-1, obs_length])
+    np.save(os.path.join(save_path, mode + "_" + str(obs_length) + "_x.npy"), data_x)
+    data_y = np.reshape(data_y, [-1, 1])
+    np.save(os.path.join(save_path, mode + "_" + str(obs_length) + "_y.npy"), data_y)
+
+
+def output_sentence(target_usage):
+    out = f"This client will consume {target_usage} kWh of electricity."
+
+    return out
+
+
+def input_sentence(usage, client_id, start_date, obs_length):
+    end_day = add_days(start_date, delta_days=obs_length - 1)
+    prediction_day = add_days(start_date, delta_days=obs_length)
+    start_week_day = datetime.datetime.strptime(start_date, '%B %d, %Y').strftime('%A')
+    end_week_day = datetime.datetime.strptime(end_day, '%B %d, %Y').strftime('%A')
+    prediction_week_day = datetime.datetime.strptime(prediction_day, '%B %d, %Y').strftime('%A')
+    num_visits_string = ', '.join(map(str, usage))
+    out = f"From {start_date}, {start_week_day} to {end_day}, {end_week_day}, client {client_id} consumed {num_visits_string} kWh of electricity on each day. What is the consumption going to be on {prediction_day}, {prediction_week_day}?"
+
+    return out
+
+
+def generate_prompt(raw_folder, save_path, mode="train", obs_length=15, first_day="January 1, 2012"):
+    raw_data = np.load(os.path.join(raw_folder, "ecl_raw_" + mode + ".npy"))
+    data_x_prompt = []
+    data_y_prompt = []
+    for i in range(len(raw_data)):
+        data = raw_data[i]
+        number_of_instance = len(data) - obs_length
+        for j in range(number_of_instance):
+            start_day = add_days(first_day, j)
+            y = data[obs_length + j]
+            x = data[j: obs_length + j]
+            data_y_prompt.append(output_sentence(y))
+            data_x_prompt.append(input_sentence(x, i+1, start_day, obs_length))
+
+    with open(os.path.join(save_path, mode + "_15_x_prompt.txt"), "w") as f:
+        for i in data_x_prompt:
+            f.write(i + "\n")
+        f.close()
+
+    with open(os.path.join(save_path, mode + "_y_prompt.txt"), "w") as f:
+        for i in data_y_prompt:
+            f.write(i + "\n")
+        f.close()