Handling Missing Values when Applying Classification Models
By Maytal Saar-Tsechansky, Foster J. Provost. 2007
388 citations recorded by the source corpus.
Subfields: Artificial Intelligence, Environmental Engineering
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
- Florian Tramèr
- Salvatore Ruggieri
- Timothy C. Au
- Ari Juels
- Thomas Ristenpart
- Didier Stricker
- Attila Reiss
- Nicolas Papernot
In Environmental Engineering
- Salvatore Ruggieri
- Timothy C. Au
- Didier Stricker
- Attila Reiss
- Erik Rodner
- Alexander Freytag
- Joachim Denzler
- Christoph Göring
Citations in the corpus
- Stealing Machine Learning Models via Prediction APIs (2016, background): “One way of handling such inputs ([11, 46]) is to label each node in the tree with an output value.”
- The Opportunity challenge: A benchmark database for on-body sensor-based activity recognition (2013, methodology): “Saar-Tsechansky and Provost (2007)), in this study we simply repeat the previous value in place of the missing value.”
- Nonparametric Part Transfer for Fine-Grained Recognition (2014, methodology): “In absence of further knowledge, we use zero imputation [29].”
- Creating and benchmarking a new dataset for physical activity monitoring (2012, methodology): “more complex methods [21] in future improved approaches.”
- Subtree Replacement in Decision Tree Simplification (2012, background): “For cases with missing value for the attribute selected at a node, several options are available [23].”
- Random Forests, Decision Trees, and Categorical Predictors: The "Absent Levels" Problem (2017, methodology): “These pseudo-instances are then sent down their appropriate daughter nodes in order to proceed down the tree as usual, and the final prediction is derived from the weighted predictions of all the terminal nodes that are subsequently reached (Saar-Tsechansky and Provost, 2007).”
- Complete Search for Feature Selection in Decision Trees (2019, methodology): “For instances with missing value of the attribute tested at a decision node, several options are available (Saar-Tsechansky and Provost, 2007; Twala, 2009).”
- Active Feature-Value Acquisition (2009, background): “The loss in predictive accuracy stemming from the need to estimate missing predictors’ values or their distributions can be avoided if the model incorporates only predictors whose values are known for this instance (Saar-Tsechansky and Provost 2007).”