데이터 사이언티스트 콧대 높았고, 연봉도 많았던 것이 사실입니다.
하지만 시대가 또 다른 국면으로 바뀌는 것 같습니다.
챗GPT ChatGPT가 코딩은 좀 합니다. 데이터 사이언스도 익숙하지요.
간혹 틀리는 것이 있더라도 이 정도 까지는 틀을 만들 수 있습니다. 문제는 어떤 식으로 시킬 것인지에 대한 [ #promptengineering ]
#프롬프트엔지니어링
인간 데이터 사이언티스트가 살아남기 위해 유일한 전환 가능한 방향이 아닐지 싶습니다.
이제 어쩔 수 없이 시대는 데이터 사이언스하기가 아니라 데이터 사이언스 시키기(prompting data science)로 바뀌는 것 같습니다.
진짜인지 예를 한번 볼까요?
ChatGPT RESPONSE:
To create a customer visit prediction model using CatBoost, first, you need to install the CatBoost library by running !pip install catboost in your terminal or Jupyter Notebook. Then, you can follow the steps below:
- Import necessary libraries
- Load the data
- Preprocess the data
- Train the model
- Evaluate the model
- Save the model (optional)
Here's a sample Python script to help you get started:
import pandas as pd
import numpy as np
from catboost import CatBoostClassifier, Pool
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, classification_report, confusion_matrix
# Load data
orders = pd.read_csv("order_transactions.csv")
clicks = pd.read_csv("click_stream.csv")
demographics = pd.read_csv("customer_demographics.csv")
# Preprocess the data
# Merge data sources into a single DataFrame
merged_data = orders.merge(clicks, on='customer_id').merge(demographics, on='customer_id')
# Feature engineering
# Add any relevant features based on the available data
# For example, calculate the total number of orders per customer, average time spent on the website, etc.
# Define target variable: whether a customer will visit the website within the next month (1) or not (0)
merged_data['next_month_visit'] = ...
# Define features and target
X = merged_data.drop('next_month_visit', axis=1)
y = merged_data['next_month_visit']
# Split the data into training and testing sets
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state=42)
# Identify categorical features
cat_features = np.where(X_train.dtypes != np.float)[0]
# Train the model
model = CatBoostClassifier(iterations=1000, learning_rate=0.1, depth=6, loss_function='Logloss', cat_features=cat_features)
model.fit(X_train, y_train, eval_set=(X_test, y_test), verbose=100)
# Evaluate the model
y_pred = model.predict(X_test)
accuracy = accuracy_score(y_test, y_pred)
print(f"Accuracy: {accuracy:.4f}")
# Print detailed evaluation
print(classification_report(y_test, y_pred))
print(confusion_matrix(y_test, y_pred))
# Save the model (optional)
model.save_model('customer_visit_prediction_model.cbm')
Replace the file names with the actual CSV file names and preprocess the data accordingly. Also, make sure to define
the target variable (next_month_visit) appropriately based on your available data.
조금더 진도를 나가 볼까요?
To add 10 new derived features, you can create new columns in the merged_data DataFrame based on the existing data. Here are 10 example features that you can create:
- Total number of orders per customer
- Average order value per customer
- Total number of clicks per customer
- Average time spent on the website per session
- Number of days since the last order
- Number of days since the last session
- Average order value in the last 30 days
- Average number of clicks in the last 30 days
- The ratio of orders to clicks
- Age group based on the customer's age
- Here's how you can add these features:
# Feature engineering
# 1. Total number of orders per customer
merged_data['total_orders'] = merged_data.groupby('customer_id')['order_id'].transform('count')
# 2. Average order value per customer
merged_data['avg_order_value'] = merged_data.groupby('customer_id')['order_value'].transform('mean')
# 3. Total number of clicks per customer
merged_data['total_clicks'] = merged_data.groupby('customer_id')['click_id'].transform('count')
# 4. Average time spent on the website per session
merged_data['avg_session_duration'] = merged_data.groupby('customer_id')['session_duration'].transform('mean')
# 5. Number of days since the last order
merged_data['days_since_last_order'] = (pd.to_datetime('today') - pd.to_datetime(merged_data['order_date'])).dt.days
# 6. Number of days since the last session
merged_data['days_since_last_session'] = (pd.to_datetime('today') - pd.to_datetime(merged_data['session_date'])).dt.days
# 7. Average order value in the last 30 days
last_30_days = merged_data[merged_data['order_date'] > (pd.to_datetime('today') - pd.Timedelta(days=30))]
merged_data['avg_order_value_30d'] = last_30_days.groupby('customer_id')['order_value'].transform('mean')
# 8. Average number of clicks in the last 30 days
last_30_days_clicks = merged_data[merged_data['session_date'] > (pd.to_datetime('today') - pd.Timedelta(days=30))]
merged_data['avg_clicks_30d'] = last_30_days_clicks.groupby('customer_id')['click_id'].transform('count')
# 9. The ratio of orders to clicks
merged_data['order_click_ratio'] = merged_data['total_orders'] / merged_data['total_clicks']
# 10. Age group based on the customer's age
age_bins = [0, 18, 24, 34, 44, 54, 64, np.inf]
age_labels = ['<18', '18-24', '25-34', '35-44', '45-54', '55-64', '65+']
merged_data['age_group'] = pd.cut(merged_data['age'], bins=age_bins, labels=age_labels)
Here are five additional features that can potentially be important and not redundant:
- Number of unique product categories ordered
- The most frequent product category ordered
- Total revenue per customer
- The proportion of sessions with a purchase
- Average number of items per order
- Here's how you can add these features:
# Feature engineering (continued)
# 1. Number of unique product categories ordered
merged_data['unique_categories'] = merged_data.groupby('customer_id')['product_category'].transform('nunique')
# 2. The most frequent product category ordered
merged_data['most_frequent_category'] = merged_data.groupby('customer_id')['product_category'].transform(lambda x: x.mode().iloc[0])
# 3. Total revenue per customer
merged_data['total_revenue'] = merged_data.groupby('customer_id')['order_value'].transform('sum')
# 4. The proportion of sessions with a purchase
merged_data['purchase_sessions'] = merged_data.groupby('customer_id')['session_id'].transform(lambda x: x[x['order_id'].notnull()].count())
merged_data['total_sessions'] = merged_data.groupby('customer_id')['session_id'].transform('count')
merged_data['purchase_session_ratio'] = merged_data['purchase_sessions'] / merged_data['total_sessions']
# 5. Average number of items per order
merged_data['avg_items_per_order'] = merged_data.groupby('customer_id')['items_count'].transform('mean')
As before, these derived features are examples, and you can create additional features or modify these based on your specific use case and data. Be sure to replace the column names with the actual column names in your dataset.
https://www.youtube.com/watch?v=8m2wLlyvPoQ&t=9s
promptStrategies
is a project brand of revisionConsulting focused on Prompt Engineering in action
Master ChatGPT Prompt Engineering with promptStrategies!
Boost Your Business Today! Unlock the full potential of ChatGPT for your business and marketing efforts with promptStrategies! Our program is designed to teach and consult on end-user level ChatGPT prompt engineering, empowering you to achieve better results. In this video, you'll discover how prompt engineering strategies can benefit your business by improving AI-generated content quality, increasing efficiency, tailoring outputs, and enhancing customer engagement and conversion rates. Hear from our satisfied customers, and learn how our experts will walk you through the limitations of current AI technologies, equipping you with practical strategies to overcome them. Don't miss this opportunity to boost your ChatGPT performance and elevate your business to new heights! Follow the facebook "PromptStrategies" page to explore promptStrategies and start your journey toward mastering ChatGPT prompt engineering today! Remember, the future of your business depends on your ability to adapt and make the most of AI advancements. Let promptStrategies be your guide. promptStrategies - Unleash the true power of ChatGPT for your business.
fb: PromptStrategies
#ChatGPT #promptStrategies #AI #BusinessGrowth #MarketingStrategy #boostyourbusiness