日常数据科学任务的 20 个有效Prompts
- Medium：https://medium.com/aimonks/
1. 数据清洗与预处理
Data Cleaning and Preprocessing:
Act as a data analyst and identify missing values, outliers, and duplicate entries in the dataset [dataset name]. Suggest strategies for handling these issues to prepare the data for analysis.
扮演数据分析师，识别数据集 [数据集名称] 中的缺失值、异常值和重复条目。提出处理这些问题的策略，为分析数据做好准备。
2. 探索性数据分析 (EDA)
Exploratory Data Analysis (EDA):
Act as a data scientist exploring a new dataset. Summarize the dataset [dataset name], including its size, structure, and the distribution of key variables. Highlight any interesting correlations or patterns observed.
扮演数据科学家探索新数据集。总结数据集 [数据集名称]，包括其规模、结构和关键变量的分布。强调观察到的任何有趣的相关性或模式。
3. 特征工程
Feature Engineering:
Act as a machine learning engineer and create new features for the dataset [dataset name] that could improve the performance of a predictive model. Describe the rationale behind each new feature.
扮演机器学习工程师，为数据集[数据集名称]创建可提高预测模型性能的新特征。描述每个新特征背后的原理。
4. 数据可视化
Data Visualization:
Act as a data visualization expert and design visualizations that effectively communicate the findings of the exploratory data analysis of [dataset name], including histograms, scatter plots, and heatmaps.
扮演数据可视化专家，设计能有效传达[数据集名称]探索性数据分析结果的可视化方法，包括直方图、散点图和热图等。
5. 统计分析
Statistical Analysis:
Act as a statistician analyzing the dataset [dataset name]. Perform hypothesis testing to determine if there are statistically significant differences between groups [group A] and [group B] with respect to [variable].
以统计学家的身份分析数据集[数据集名称]。进行假设检验，以确定[A 组]和[B 组]在[变量]方面是否存在显著的统计学差异。
6. 模型选择
Model Selection:
Act as a data science consultant tasked with choosing the best machine learning model for predicting [target variable] based on the features in [dataset name]. Discuss the pros and cons of at least three models.
担任数据科学顾问，根据[数据集名称]中的特征选择预测[目标变量]的最佳机器学习模型。讨论至少三种模型的优缺点。
7. 模型训练
Model Training:
Act as a machine learning specialist training a model on [dataset name]. Outline the steps for preprocessing data, splitting it into training and test sets, and training a [model type] model.
扮演机器学习专家，在[数据集名称]上训练模型。概述预处理数据、将数据分成训练集和测试集以及训练[模型类型]模型的步骤。
8. 模型评估
Model Evaluation:
Act as a machine learning evaluator assessing the performance of a [model type] on the test set of [dataset name]. Calculate accuracy, precision, recall, and F1 score, and interpret these metrics.
担任机器学习评估员，评估[模型类型]在[数据集名称]测试集上的性能。计算准确率、精确率、召回率和 F1 分数，并解释这些指标。
9. 超参数调整
Hyperparameter Tuning:
Act as a machine learning engineer optimizing a [model type] for dataset [dataset name]. Describe a strategy for hyperparameter tuning, including the selection of parameters to adjust and the tuning method.
扮演机器学习工程师，针对数据集[数据集名称]优化[模型类型]。描述超参数调整策略，包括选择要调整的参数和调整方法。
10. 交叉验证
Cross-Validation:
Act as a data analyst performing cross-validation on [dataset name] using a [model type]. Explain the cross-validation process and how it helps in assessing the model’s generalizability.
扮演数据分析师，使用[模型类型]对[数据集名称]进行交叉验证。解释交叉验证过程以及它如何有助于评估模型的普适性。
11. 预测分析
Predictive Analysis:
Act as a predictive analyst making forecasts using the trained [model type] on dataset [dataset name]. Provide predictions for the next [time period] and discuss the confidence in these predictions.
扮演预测分析师，使用训练有素的[模型类型]对数据集[数据集名称]进行预测。提供下一个[时间段]的预测，并讨论对这些预测的置信度。
12. 文本分析
Text Analysis:
Act as a NLP specialist analyzing text data in [dataset name]. Conduct sentiment analysis, topic modeling, and keyword extraction to uncover insights from text data.
担任 NLP 专家，分析[数据集名称]中的文本数据。进行情感分析、主题建模和关键字提取，从文本数据中挖掘洞察力。
13. 时间序列分析
Time Series Analysis:
Act as a time series analyst working with dataset [dataset name]. Apply ARIMA/SARIMA models to forecast [target variable] for the next [time period] and discuss the model’s assumptions and confidence intervals.
担任时间序列分析员，处理数据集[数据集名称]。应用 ARIMA/SARIMA 模型预测下一个[时间段]的[目标变量]，并讨论模型的假设和置信区间。
14. 异常检测
Anomaly Detection:
Act as a data scientist, identifying anomalies in [dataset name]. Use [specified method] to detect outliers and potential anomalies in the dataset and suggest possible explanations for these anomalies.
扮演数据科学家，识别[数据集名称]中的异常情况。使用[指定方法]检测数据集中的异常值和潜在异常，并对这些异常提出可能的解释。
15. 聚类分析
Cluster Analysis:
Act as a machine learning engineer performing cluster analysis on dataset [dataset name]. Identify natural groupings in the data using the [specified clustering technique] and interpret the characteristics of each cluster.
扮演机器学习工程师，对数据集[数据集名称]进行聚类分析。使用[指定聚类技术]识别数据中的自然分组，并解释每个聚类的特征。
16. 降维
Dimensionality Reduction:
Act as a data analyst applying dimensionality reduction to dataset [dataset name] using PCA/TSNE. Explain the process and how it aids in visualization and model performance.
担任数据分析师，使用 PCA/TSNE 对数据集[数据集名称]进行降维处理。解释这一过程以及它如何有助于可视化和模型性能。
17. 数据整合
Data Integration:
Act as a data engineer, integrating multiple datasets for a comprehensive analysis. Describe the process of combining [dataset A] and [dataset B], including handling inconsistencies and ensuring data quality.
扮演数据工程师，整合多个数据集进行综合分析。描述[数据集 A]和[数据集 B]的组合过程，包括处理不一致之处和确保数据质量。
18. 数据管道自动化
Automating Data Pipelines:
Act as a data engineer, designing an automated pipeline for data processing and analysis of [dataset name]. Outline the components of the pipeline, including data ingestion, cleaning, transformation, and storage.
扮演数据工程师，为[数据集名称]的数据处理和分析设计自动化管道。概述管道的组成部分，包括数据获取、清理、转换和存储。
19. 部署机器学习模型
Deploying Machine Learning Models:
Act as a machine learning engineer deploying a [model type] model trained on [dataset name]. Describe the steps for deployment, including model serialization, creating a prediction API, and monitoring model performance.
扮演机器学习工程师，部署在[数据集名称]上训练的[模型类型]模型。描述部署步骤，包括模型序列化、创建预测 API 和监控模型性能。
20. 数据科学中的伦理问题
Ethical Considerations in Data Science:
Act as an ethicist, addressing ethical considerations in data science projects. Discuss how to handle sensitive data in [dataset name], including privacy, bias, and fairness, and suggest best practices for ethical data science.
扮演伦理学家，解决数据科学项目中的伦理问题。讨论如何处理[数据集名称]中的敏感数据，包括隐私、偏见和公平性，并提出符合伦理的数据科学最佳实践。
