Export
数据分析智能体：瞬间获取数据洞察 ✨
本教程为高级教程。建议先了解另一本手册的内容！
在本 Notebook 中，我们将创建一个数据分析智能体：一个配备数据分析库的代码智能体，能够加载和转换数据框，从中提取洞察，甚至绘制结果图表！
假设我想分析 Kaggle Titanic 挑战的数据，以预测每个乘客的生还情况。但在我深入挖掘之前，我希望一个自动化智能体为我准备分析，提取趋势并绘制一些图形来寻找洞察。
让我们开始设置这个系统。
运行下面的代码以安装所需的依赖：
[ ]
我们首先创建智能体。我们使用了 ReactCodeAgent（请阅读文档了解更多关于智能体类型的信息），因此我们甚至不需要为其提供任何工具：它可以直接运行代码。
我们只需要确保它能够使用与数据科学相关的库，方法是将这些库传递给 additional_authorized_imports 参数：["numpy", "pandas", "matplotlib.pyplot", "seaborn"]。
一般来说，当在 additional_authorized_imports 中传递库时，确保这些库已在本地环境中安装，因为 Python 解释器只能使用已安装的库。
⚙ 我们的智能体将由 meta-llama/Meta-Llama-3.1-70B-Instruct 提供支持，使用 HfEngine 类，这个类通过 HF 的推理 API 实现：推理 API 使得运行任何操作系统模型变得快速而简单。
[ ]
数据分析 📊🤔
在运行智能体时，我们提供了来自竞赛的额外说明，并将其作为关键字参数（kwarg）传递给 run 方法：
[4]
[ ]
[6]
Here are the correlations and trends found in the data: 1. **Correlation between age and survival rate**: The correlation is -0.0772, which suggests that as age increases, the survival rate decreases. This implies that older passengers were less likely to survive the Titanic disaster. 2. **Relationship between Pclass and survival rate**: The survival rates for each Pclass are: - Pclass 1: 62.96% - Pclass 2: 47.28% - Pclass 3: 24.24% This shows that passengers in higher socio-economic classes (Pclass 1 and 2) had a significantly higher survival rate compared to those in the lower class (Pclass 3). 3. **Relationship between fare and survival rate**: The correlation is 0.2573, which suggests a moderate positive relationship between fare and survival rate. This implies that passengers who paid higher fares were more likely to survive the disaster.
令人印象深刻，不是吗？你还可以为你的智能体提供一个可视化工具，让它能够反思自己绘制的图表！
数据科学智能体：进行预测 🛠️
👉 现在让我们深入一步：我们将让我们的模型在数据上执行预测。
为此，我们还需要让它使用 sklearn，并将其添加到 additional_authorized_imports 中。
[8]
======== New task ======== You are an expert machine learning engineer. Please train a ML model on "titanic/train.csv" to predict the survival for rows of "titanic/test.csv". Output the results under './output.csv'. Take care to import functions and modules before using them! You have been provided with these initial arguments: {'additional_notes': "\n### Variable Notes\npclass: A proxy for socio-economic status (SES)\n1st = Upper\n2nd = Middle\n3rd = Lower\nage: Age is fractional if less than 1. If the age is estimated, is it in the form of xx.5\nsibsp: The dataset defines family relations in this way...\nSibling = brother, sister, stepbrother, stepsister\nSpouse = husband, wife (mistresses and fiancés were ignored)\nparch: The dataset defines family relations in this way...\nParent = mother, father\nChild = daughter, son, stepdaughter, stepson\nSome children travelled only with a nanny, therefore parch=0 for them.\n\nHere is the final answer:\n\n**Interesting Questions**\n\n1. Is there a correlation between socio-economic status (Pclass) and survival rate?\n2. Is there a correlation between age and survival rate?\n3. Is there a correlation between family size and survival rate?\n\n**Answers to Questions**\n\n1. Correlation between Pclass and survival rate: -0.338\nInsight: Passengers from lower socio-economic backgrounds were less likely to survive.\n2. Correlation between age and survival rate: -0.077\nInsight: Older passengers were slightly less likely to survive.\n3. Correlation between family size and survival rate: 0.017\nInsight: Passengers traveling with larger families were slightly more likely to survive, but this correlation is very weak and may not be significant.\n\n**Summary of Correlations and Trends**\n\nThe analysis of the Titanic dataset reveals several interesting trends and correlations. Firstly, the socio-economic status of passengers played a significant role in their survival, with passengers from lower socio-economic backgrounds being less likely to survive. Secondly, age was a weak predictor of survival, with older passengers being slightly less likely to survive. Finally, family size had a very weak positive correlation with survival rate, suggesting that passengers traveling with larger families were slightly more likely to survive, but this correlation is very weak and may not be significant.\n\n**Plots**\n\n(Attached are the plots generated using matplotlib/seaborn and saved to the './figures/' folder)\n\nI hope this meets the requirements!"}. ==== Agent is executing the code below: import pandas as pd from sklearn.model_selection import train_test_split from sklearn.ensemble import RandomForestClassifier from sklearn.metrics import accuracy_score # Load the dataset train_df = pd.read_csv('titanic/train.csv') test_df = pd.read_csv('titanic/test.csv') print(train_df.head()) print(test_df.head()) ==== Print outputs: PassengerId Survived Pclass \ 0 1 0 3 1 2 1 1 2 3 1 3 3 4 1 1 4 5 0 3 Name Sex Age SibSp \ 0 Braund, Mr. Owen Harris male 22.0 1 1 Cumings, Mrs. John Bradley (Florence Briggs Th... female 38.0 1 2 Heikkinen, Miss. Laina female 26.0 0 3 Futrelle, Mrs. Jacques Heath (Lily May Peel) female 35.0 1 4 Allen, Mr. William Henry male 35.0 0 Parch Ticket Fare Cabin Embarked 0 0 A/5 21171 7.2500 NaN S 1 0 PC 17599 71.2833 C85 C 2 0 STON/O2. 3101282 7.9250 NaN S 3 0 113803 53.1000 C123 S 4 0 373450 8.0500 NaN S PassengerId Pclass Name Sex \ 0 892 3 Kelly, Mr. James male 1 893 3 Wilkes, Mrs. James (Ellen Needs) female 2 894 2 Myles, Mr. Thomas Francis male 3 895 3 Wirz, Mr. Albert male 4 896 3 Hirvonen, Mrs. Alexander (Helga E Lindqvist) female Age SibSp Parch Ticket Fare Cabin Embarked 0 34.5 0 0 330911 7.8292 NaN Q 1 47.0 1 0 363272 7.0000 NaN S 2 62.0 0 0 240276 9.6875 NaN Q 3 27.0 0 0 315154 8.6625 NaN S 4 22.0 1 1 3101298 12.2875 NaN S ==== Agent is executing the code below: # Handle missing values train_df['Age'].fillna(train_df['Age'].median(), inplace=True) test_df['Age'].fillna(test_df['Age'].median(), inplace=True) train_df['Cabin'].fillna('Unknown', inplace=True) test_df['Cabin'].fillna('Unknown', inplace=True) # Encode categorical variables train_df['Sex'] = train_df['Sex'].map({'male': 0, 'female': 1}) test_df['Sex'] = test_df['Sex'].map({'male': 0, 'female': 1}) train_df['Embarked'] = train_df['Embarked'].fillna('S') test_df['Embarked'] = test_df['Embarked'].fillna('S') train_df['Embarked'] = train_df['Embarked'].map({'S': 0, 'C': 1, 'Q': 2}) test_df['Embarked'] = test_df['Embarked'].map({'S': 0, 'C': 1, 'Q': 2}) print(train_df.head()) print(test_df.head()) ==== Print outputs: PassengerId Survived Pclass \ 0 1 0 3 1 2 1 1 2 3 1 3 3 4 1 1 4 5 0 3 Name Sex Age SibSp Parch \ 0 Braund, Mr. Owen Harris 0 22.0 1 0 1 Cumings, Mrs. John Bradley (Florence Briggs Th... 1 38.0 1 0 2 Heikkinen, Miss. Laina 1 26.0 0 0 3 Futrelle, Mrs. Jacques Heath (Lily May Peel) 1 35.0 1 0 4 Allen, Mr. William Henry 0 35.0 0 0 Ticket Fare Cabin Embarked 0 A/5 21171 7.2500 Unknown 0 1 PC 17599 71.2833 C85 1 2 STON/O2. 3101282 7.9250 Unknown 0 3 113803 53.1000 C123 0 4 373450 8.0500 Unknown 0 PassengerId Pclass Name Sex \ 0 892 3 Kelly, Mr. James 0 1 893 3 Wilkes, Mrs. James (Ellen Needs) 1 2 894 2 Myles, Mr. Thomas Francis 0 3 895 3 Wirz, Mr. Albert 0 4 896 3 Hirvonen, Mrs. Alexander (Helga E Lindqvist) 1 Age SibSp Parch Ticket Fare Cabin Embarked 0 34.5 0 0 330911 7.8292 Unknown 2 1 47.0 1 0 363272 7.0000 Unknown 0 2 62.0 0 0 240276 9.6875 Unknown 2 3 27.0 0 0 315154 8.6625 Unknown 0 4 22.0 1 1 3101298 12.2875 Unknown 0 ==== Agent is executing the code below: # Split data into features (X) and target (y) X = train_df.drop(['Survived', 'PassengerId', 'Name', 'Ticket'], axis=1) y = train_df['Survived'] # Split data into training and validation sets X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42) # Train a Random Forest Classifier model rfc = RandomForestClassifier(n_estimators=100, random_state=42) rfc.fit(X_train, y_train) print("Training accuracy:", accuracy_score(y_train, rfc.predict(X_train))) print("Validation accuracy:", accuracy_score(y_val, rfc.predict(X_val))) ==== Code execution failed due to the following error: could not convert string to float: 'C124' Traceback (most recent call last): File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 1054, in step result = self.python_evaluator( File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 893, in evaluate_python_code result = evaluate_ast(node, state, static_tools, custom_tools, authorized_imports) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 780, in evaluate_ast return evaluate_ast(expression.value, state, static_tools, custom_tools) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 744, in evaluate_ast return evaluate_call(expression, state, static_tools, custom_tools) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 439, in evaluate_call output = func(*args, **kwargs) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/base.py", line 1474, in wrapper return fit_method(estimator, *args, **kwargs) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/ensemble/_forest.py", line 363, in fit X, y = self._validate_data( File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/base.py", line 650, in _validate_data X, y = check_X_y(X, y, **check_params) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/utils/validation.py", line 1263, in check_X_y X = check_array( File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/utils/validation.py", line 997, in check_array array = _asarray_with_order(array, order=order, dtype=dtype, xp=xp) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/utils/_array_api.py", line 521, in _asarray_with_order array = numpy.asarray(array, order=order, dtype=dtype) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/pandas/core/generic.py", line 2153, in __array__ arr = np.asarray(values, dtype=dtype) ValueError: could not convert string to float: 'C124' During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 756, in direct_run step_logs = self.step() File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 1072, in step raise AgentExecutionError(error_msg) transformers.agents.agents.AgentExecutionError: Code execution failed due to the following error: could not convert string to float: 'C124' ==== Agent is executing the code below: # One-hot encode the Cabin feature cabin_dummies = pd.get_dummies(train_df['Cabin']) test_cabin_dummies = pd.get_dummies(test_df['Cabin']) X = pd.concat([train_df.drop(['Survived', 'PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1), cabin_dummies], axis=1) test_X = pd.concat([test_df.drop(['PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1), test_cabin_dummies], axis=1) y = train_df['Survived'] # Split data into training and validation sets X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42) # Train a Random Forest Classifier model rfc = RandomForestClassifier(n_estimators=100, random_state=42) rfc.fit(X_train, y_train) print("Training accuracy:", accuracy_score(y_train, rfc.predict(X_train))) print("Validation accuracy:", accuracy_score(y_val, rfc.predict(X_val))) ==== Print outputs: Training accuracy: 0.9845505617977528 Validation accuracy: 0.7932960893854749 ==== Agent is executing the code below: # Make predictions on the test data predictions = rfc.predict(test_X) # Save the predictions to a submission file submission_df = pd.DataFrame({ 'PassengerId': test_df['PassengerId'], 'Survived': predictions }) submission_df.to_csv('./output.csv', index=False) final_answer("Submission file saved to./output.csv") ==== Code execution failed due to the following error: The feature names should match those that were passed during fit. Feature names unseen at fit time: - A11 - A18 - A21 - A29 - A9 - ... Feature names seen at fit time, yet now missing: - A10 - A14 - A16 - A19 - A20 - ... Traceback (most recent call last): File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 1054, in step result = self.python_evaluator( File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 893, in evaluate_python_code result = evaluate_ast(node, state, static_tools, custom_tools, authorized_imports) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 739, in evaluate_ast return evaluate_assign(expression, state, static_tools, custom_tools) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 322, in evaluate_assign result = evaluate_ast(assign.value, state, static_tools, custom_tools) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 744, in evaluate_ast return evaluate_call(expression, state, static_tools, custom_tools) File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/python_interpreter.py", line 439, in evaluate_call output = func(*args, **kwargs) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/ensemble/_forest.py", line 905, in predict proba = self.predict_proba(X) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/ensemble/_forest.py", line 947, in predict_proba X = self._validate_X_predict(X) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/ensemble/_forest.py", line 641, in _validate_X_predict X = self._validate_data( File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/base.py", line 608, in _validate_data self._check_feature_names(X, reset=reset) File "/Users/aymeric/venvs/disposable/lib/python3.10/site-packages/sklearn/base.py", line 535, in _check_feature_names raise ValueError(message) ValueError: The feature names should match those that were passed during fit. Feature names unseen at fit time: - A11 - A18 - A21 - A29 - A9 - ... Feature names seen at fit time, yet now missing: - A10 - A14 - A16 - A19 - A20 - ... During handling of the above exception, another exception occurred: Traceback (most recent call last): File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 756, in direct_run step_logs = self.step() File "/Users/aymeric/Documents/Code/original_transformers/transformers/src/transformers/agents/agents.py", line 1072, in step raise AgentExecutionError(error_msg) transformers.agents.agents.AgentExecutionError: Code execution failed due to the following error: The feature names should match those that were passed during fit. Feature names unseen at fit time: - A11 - A18 - A21 - A29 - A9 - ... Feature names seen at fit time, yet now missing: - A10 - A14 - A16 - A19 - A20 - ... ==== Agent is executing the code below: # Get the common cabin values in both training and test data common_cabins = set(train_df['Cabin'].unique()) & set(test_df['Cabin'].unique()) # Filter the cabin dummies to only include the common cabin values cabin_dummies = pd.get_dummies(train_df['Cabin'].map(lambda x: x if x in common_cabins else 'Unknown')) test_cabin_dummies = pd.get_dummies(test_df['Cabin'].map(lambda x: x if x in common_cabins else 'Unknown')) X = pd.concat([train_df.drop(['Survived', 'PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1), cabin_dummies], axis=1) test_X = pd.concat([test_df.drop(['PassengerId', 'Name', 'Ticket', 'Cabin'], axis=1), test_cabin_dummies], axis=1) y = train_df['Survived'] # Split data into training and validation sets X_train, X_val, y_train, y_val = train_test_split(X, y, test_size=0.2, random_state=42) # Train a Random Forest Classifier model rfc = RandomForestClassifier(n_estimators=100, random_state=42) rfc.fit(X_train, y_train) print("Training accuracy:", accuracy_score(y_train, rfc.predict(X_train))) print("Validation accuracy:", accuracy_score(y_val, rfc.predict(X_val))) # Make predictions on the test data predictions = rfc.predict(test_X) # Save the predictions to a submission file submission_df = pd.DataFrame({ 'PassengerId': test_df['PassengerId'], 'Survived': predictions }) submission_df.to_csv('./output.csv', index=False) final_answer("Submission file saved to./output.csv") ==== Print outputs: Training accuracy: 0.9803370786516854 Validation accuracy: 0.8100558659217877 >>> Final answer: Submission file saved to./output.csv
智能体输出的测试预测，一旦提交到 Kaggle，得分为 0.78229，在 17,360 名参赛者中排名 #2824，而且比我几年前第一次尝试这个挑战时艰难取得的成绩还要好。
你的结果可能会有所不同，但无论如何，我认为能够在几秒钟内通过智能体实现这一点，实在是非常令人印象深刻。
🚀 以上只是一个数据分析智能体的简单尝试：它肯定可以在很多方面进行改进，以更好地适应你的具体使用场景！