Data
This digit database was created by collecting 250 samples from 44 writers. WACOM PL-100V pressure sensitive tablet used with an integrated LCD display and a cordless stylus. The input and display areas are located in the same place. Attached to the serial port of an Intel 486 based PC, it allows us to collect handwriting samples. The tablet sends $x$ and $y$ tablet coordinates and pressure level values of the pen at fixed time intervals (sampling rate) of 100 miliseconds.
These writers are asked to write 250 digits in random order inside boxes of 500 by 500 tablet pixel resolution. Subject are monitored only during the first entry screens. Each screen contains five boxes with the digits to be written displayed above. Subjects are told to write only inside these boxes. If they make a mistake or are unhappy with their writing, they are instructed to clear the content of a box by using an on-screen button. The first ten digits are ignored because most writers are not familiar with this type of input devices, but subjects are not aware of this.
In this study, researchers use only ($x, y$) coordinate information. The stylus pressure level values are ignored. The raw data that we capture from the tablet consist of integer values between 0 and 500 (tablet input box resolution). The new coordinates are such that the coordinate which has the maximum range varies between 0 and 100. Usually $x$ stays in this range, since most characters are taller than they are wide.
Attribute information:
In order to train and test our classifiers, we need to represent digits as constant length feature vectors. A commonly used technique leading to good results is resampling the ( x_t, y_t) points. Temporal resampling (points regularly spaced in time) or spatial resampling (points regularly spaced in arc length) can be used here. Raw point data are already regularly spaced in time but the distance between them is variable.
- Data reference link :
https://archive.ics.uci.edu/ml/datasets/Pen-Based+Recognition+of+Handwritten+Digits
Boxplot
Outliers
Data (input2, input4, and input13) have outliers. All rows which contain outliers will be dropped. However, both data will be used for models in order to compare the results.
Comparison between cleaned and uncleaned data results
* Cleaned data (with dropped outliers) is slightly better than uncleaned data.
* Outlier cleaning seemed to help clean the noise in the data.
* The current results are good enough. But for better results, further cleaning can be done considering the class outliers individually.
Comparison between cleaned, uncleaned, bagging results
* Decision Tree model test results were slightly increased while train results decreased with cleaned data. It seems that overfit decreased slightly, after dropping outliers.
* Cleaned data also used with bagging method for comparison. The results were better both for test and training scores. Overall results showed that overfitting eliminated with a better test scores with cleaned data & bagging.
Results
* All four models has pretty good scores.
* SVM has the first place in terms of scores.
* However, in the real world scenario, score not the only thing to consider.
* For example, if there is a need of explainable model, DT is the only option.
* In this context, which algorithm to choose for deployment is the subject of the business problem we plan to solve.
