AlphaClean: Automatic Generation of Data Cleaning Pipelines

By Sanjay Krishnan, Eugene Wu. 2019

61 citations recorded by the source corpus.

The analyst effort in data cleaning is gradually shifting away from the design of hand-written scripts to building and tuning complex pipelines of automated data cleaning libraries. Hyper-parameter tuning for data cleaning is very different than hyper-parameter tuning for machine learning since the pipeline components and objective functions have structure that tuning algorithms can exploit. This paper proposes a framework, called AlphaClean, that rethinks parameter tuning for data cleaning pipelines. AlphaClean provides users with a rich library to define data quality measures with weighted sums of SQL aggregate queries. AlphaClean applies generate-then-search framework where each pipelined cleaning operator contributes candidate transformations to a shared pool. Asynchronously, in separate threads, a search algorithm sequences them into cleaning pipelines that maximize the user-defined quality measures. This architecture allows AlphaClean to apply a number of optimizations including incremental evaluation of the quality measures and learning dynamic pruning rules to reduce the search space. Our experiments on real and synthetic benchmarks suggest that AlphaClean finds solutions of up-to 9x higher quality than naively applying state-of-the-art parameter tuning methods, is significantly more robust to straggling data cleaning methods and redundancy in the data cleaning library, and can incorporate state-of-the-art cleaning systems such as HoloClean as cleaning operators.

Subfields: Artificial Intelligence, Computer Networks and Communications, Management Science and Operations Research, ML Systems & Hardware

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

1. Marco F. Huber
2. Marc-André Zöller
3. Yue Zhao
4. Xin He
5. Mark Harman
6. Kaiyong Zhao
7. Lei Ma
8. Yang Liu

In Computer Networks and Communications

1. Marc-André Zöller
2. Xin He
3. Mark Harman
4. Marco F. Huber
5. Kaiyong Zhao
6. Lei Ma
7. Yang Liu
8. Jie M. Zhang

In Management Science and Operations Research

1. Marc-André Zöller
2. Xin He
3. Yue Zhao
4. Marco F. Huber
5. Kaiyong Zhao
6. Yang Liu
7. Xiaowen Chu
8. Zheng Li

In ML Systems & Hardware

1. Marc-André Zöller
2. Mark Harman
3. Marco F. Huber
4. Yang Liu

Citations in the corpus

• AutoML: A Survey of the State-of-the-Art (2019, background): “As an alternative, some studies propose possible automated data-cleaning processes [78, 79, 80, 81].”
• Benchmark and Survey of Automated Machine Learning Frameworks (2019, background): “Krishnan and Wu (2019) proposed an automatic data cleaning procedure with minimal human interaction: based on a human defined data quality function, data cleaning is treated similarly to pipeline structure search.”
• Data-centric Artificial Intelligence: A Survey (2023, unknown): “Although learning-based methods can search for the optimal preparation strategy automatically [122, 126], it remains a challenge to design the appropriate search space, and the search often requires a non-trivial amount of time.”
• PyOD: A Python Toolbox for Scalable Outlier Detection (2019, background): “PyOD has been used in various academic and commercial projects (Zhao and Hryniewicki, 2018a; Ramakrishnan et al., 2019; Krishnan and Wu, 2019).”
• Machine Learning Testing: Survey, Landscapes and Horizons (2019, background): “Krishnan and Wu [228] also targeted automatic data cleaning and proposed AlphaClean.”