Instructions to run Code:
https://seer.cancer.gov/data/
Request and download Seer*Stat.
Data is taken from Seer*Stat 8.42.

Once logged in, File -> New -> Case Listing Session

In the data, use Incidence - Seer Reseearch Data, 17 Registries, Nov 2022 Sub (2000-2020)
In the Selection, edit and add the three selection statements:
The below statements are formatted in: {Section.Variable} = <Chosen Values>
	{Therapy.RX Summ-Surg/Rad Seq} = 'No radiation and/or cancer-directed surgery','Radiation prior to surgery','Radiation after surgery'
	AND {Race, Sex, Year Dx.Year of diagnosis = '2020'
	OR {Therapy.RX Summ-Surg/Rad Seq} = 'Radiation before and after surgery','Intraoperative radiation','Intraoperative rad with other rad before/after surgery','Surgery both before and after radiation','Surgery both before and after radiation'

In the table choose columns:
RX Summ-Surg/Rad Seq
Age recode with <1 year olds
Sex
Race and origin recode (NHW, NHB, NHAIAN, NHAPI, Hispanic)
Months from diagnosis to treatment
TNM 7/CS v0204+ Schema recode
Site recode ICD-O-3 2023 Revision
Site recode ICD-O-3 2023 Revision Expanded
CS Schema - AJCC 6th Edition
Primary Site - labeled
Primary Site
Histologic Type ICD-O-3
ICD-O-3 Hist/behav
ICD-O-3 Hist/behav, malignant
Site recode ICD-O-3/WHO 2008 (for SIRs)
EOD Schema ID Recode (2010+)
Site recode - rare tumors
Tumor Size Summary (2016+)
Regional nodes positive (1988+)
Site - mal+ins (most detail)
Site - malignant (most detail)

In the sort:
RX Summ-Surg/Rad Seq
Age recode with <1 year olds

In Output put any title.
Session -> Execute

Once the matrix opens click on it. There should be 434900 occurrences.
At the navigation bar, go to Matrix -> Export -> Results as Text File..
In the new box, click CSV defaults. Name the data file as 2020-extra_removed.csv

From here we start python.
We need to import the important libraries.
In command line: pip install -r requirements.txt

Once complete, running python preprocess.py, will give two new csv files: 2020-extra_removed_preprocessed.csv and 2020-extra_removed_preprocessed_target_binary.csv

In the imblearn-classifiers-binary.py, choose if you want to run EasyEnsembleClassifier or BalancedRandomForestClassifier by changing the name variable to eec or randomforest respectively.
In the mlp-binary.py, choose which hidden_layer you want to see by changing the hidden_layer variable.
Please note that mlp-binary.py will fail to run classification reports if the class is not predicted correctly.

If you wish to run non binary, the preprocess.py needs to have lines 52-58 commented out.
The targets y file should not have binary tagged at the end.
Run mlp.py

Training your own model:
The models are provided in .joblib format and loaded as the classifiers if they already exist. To train, please remove those files. Note that large training can take up to 5 hours.