/* This program uses a neural network to predict survival(or death) within a five years span, for patients who underwent breast cancer surgery. I reached consistently accuracy above 70% on the test set, which is significant/meaningful of a correlation between survival and the dataset features. Author:Paul Caron Date:May 28, 2019 */ /*This is info about the dataset*/ /* 1. Haberman, S. J. (1976). Generalized Residuals for Log-Linear Models, Proceedings of the 9th International Biometrics Conference, Boston, pp. 104-122. 2. Landwehr, J. M., Pregibon, D., and Shoemaker, A. C. (1984), Graphical Models for Assessing Logistic Regression Models (with discussion), Journal of the American Statistical Association 79: 61-83. 3. Lo, W.-D. (1993). Logistic Regression Trees, PhD thesis, Department of Statistics, University of Wisconsin, Madison, WI. 4. Relevant Information: The dataset contains cases from a study that was conducted between 1958 and 1970 at the University of Chicago's Billings Hospital on the survival of patients who had undergone surgery for breast cancer. 5. Number of Instances: 306 6. Number of Attributes: 4 (including the class attribute) 7. Attribute Information: 1. Age of patient at time of operation (numerical) 2. Patient's year of operation (year - 1900, numerical) 3. Number of positive axillary nodes detected (numerical) 4. Survival status (class attribute) 1 = the patient survived 5 years or longer 2 = the patient died within 5 year 8. Missing Attribute Values: None */ const rawData=`30,64,1,1 30,62,3,1 30,65,0,1 31,59,2,1 31,65,4,1 33,58,10,1 33,60,0,1 34,59,0,2 34,66,9,2 34,58,30,1 34,60,1,1 34,61,10,1 34,67,7,1 34,60,0,1 35,64,13,1 35,63,0,1 36,60,1,1 36,69,0,1 37,60,0,1 37,63,0,1 37,58,0,1 37,59,6,1 37,60,15,1 37,63,0,1 38,69,21,2 38,59,2,1 38,60,0,1 38,60,0,1 38,62,3,1 38,64,1,1 38,66,0,1 38,66,11,1 38,60,1,1 38,67,5,1 39,66,0,2 39,63,0,1 39,67,0,1 39,58,0,1 39,59,2,1 39,63,4,1 40,58,2,1 40,58,0,1 40,65,0,1 41,60,23,2 41,64,0,2 41,67,0,2 41,58,0,1 41,59,8,1 41,59,0,1 41,64,0,1 41,69,8,1 41,65,0,1 41,65,0,1 42,69,1,2 42,59,0,2 42,58,0,1 42,60,1,1 42,59,2,1 42,61,4,1 42,62,20,1 42,65,0,1 42,63,1,1 43,58,52,2 43,59,2,2 43,64,0,2 43,64,0,2 43,63,14,1 43,64,2,1 43,64,3,1 43,60,0,1 43,63,2,1 43,65,0,1 43,66,4,1 44,64,6,2 44,58,9,2 44,63,19,2 44,61,0,1 44,63,1,1 44,61,0,1 44,67,16,1 45,65,6,2 45,66,0,2 45,67,1,2 45,60,0,1 45,67,0,1 45,59,14,1 45,64,0,1 45,68,0,1 45,67,1,1 46,58,2,2 46,69,3,2 46,62,5,2 46,65,20,2 46,62,0,1 46,58,3,1 46,63,0,1 47,63,23,2 47,62,0,2 47,65,0,2 47,61,0,1 47,63,6,1 47,66,0,1 47,67,0,1 47,58,3,1 47,60,4,1 47,68,4,1 47,66,12,1 48,58,11,2 48,58,11,2 48,67,7,2 48,61,8,1 48,62,2,1 48,64,0,1 48,66,0,1 49,63,0,2 49,64,10,2 49,61,1,1 49,62,0,1 49,66,0,1 49,60,1,1 49,62,1,1 49,63,3,1 49,61,0,1 49,67,1,1 50,63,13,2 50,64,0,2 50,59,0,1 50,61,6,1 50,61,0,1 50,63,1,1 50,58,1,1 50,59,2,1 50,61,0,1 50,64,0,1 50,65,4,1 50,66,1,1 51,59,13,2 51,59,3,2 51,64,7,1 51,59,1,1 51,65,0,1 51,66,1,1 52,69,3,2 52,59,2,2 52,62,3,2 52,66,4,2 52,61,0,1 52,63,4,1 52,69,0,1 52,60,4,1 52,60,5,1 52,62,0,1 52,62,1,1 52,64,0,1 52,65,0,1 52,68,0,1 53,58,4,2 53,65,1,2 53,59,3,2 53,60,9,2 53,63,24,2 53,65,12,2 53,58,1,1 53,60,1,1 53,60,2,1 53,61,1,1 53,63,0,1 54,60,11,2 54,65,23,2 54,65,5,2 54,68,7,2 54,59,7,1 54,60,3,1 54,66,0,1 54,67,46,1 54,62,0,1 54,69,7,1 54,63,19,1 54,58,1,1 54,62,0,1 55,63,6,2 55,68,15,2 55,58,1,1 55,58,0,1 55,58,1,1 55,66,18,1 55,66,0,1 55,69,3,1 55,69,22,1 55,67,1,1 56,65,9,2 56,66,3,2 56,60,0,1 56,66,2,1 56,66,1,1 56,67,0,1 56,60,0,1 57,61,5,2 57,62,14,2 57,64,1,2 57,64,9,1 57,69,0,1 57,61,0,1 57,62,0,1 57,63,0,1 57,64,0,1 57,64,0,1 57,67,0,1 58,59,0,1 58,60,3,1 58,61,1,1 58,67,0,1 58,58,0,1 58,58,3,1 58,61,2,1 59,62,35,2 59,60,0,1 59,63,0,1 59,64,1,1 59,64,4,1 59,64,0,1 59,64,7,1 59,67,3,1 60,59,17,2 60,65,0,2 60,61,1,1 60,67,2,1 60,61,25,1 60,64,0,1 61,62,5,2 61,65,0,2 61,68,1,2 61,59,0,1 61,59,0,1 61,64,0,1 61,65,8,1 61,68,0,1 61,59,0,1 62,59,13,2 62,58,0,2 62,65,19,2 62,62,6,1 62,66,0,1 62,66,0,1 62,58,0,1 63,60,1,2 63,61,0,1 63,62,0,1 63,63,0,1 63,63,0,1 63,66,0,1 63,61,9,1 63,61,28,1 64,58,0,1 64,65,22,1 64,66,0,1 64,61,0,1 64,68,0,1 65,58,0,2 65,61,2,2 65,62,22,2 65,66,15,2 65,58,0,1 65,64,0,1 65,67,0,1 65,59,2,1 65,64,0,1 65,67,1,1 66,58,0,2 66,61,13,2 66,58,0,1 66,58,1,1 66,68,0,1 67,64,8,2 67,63,1,2 67,66,0,1 67,66,0,1 67,61,0,1 67,65,0,1 68,67,0,1 68,68,0,1 69,67,8,2 69,60,0,1 69,65,0,1 69,66,0,1 70,58,0,2 70,58,4,2 70,66,14,1 70,67,0,1 70,68,0,1 70,59,8,1 70,63,0,1 71,68,2,1 72,63,0,2 72,58,0,1 72,64,0,1 72,67,3,1 73,62,0,1 73,68,0,1 74,65,3,2 74,63,0,1 75,62,1,1 76,67,0,1 77,65,3,1 78,65,1,2 83,58,2,2`; const survivalCode=[ [1,0], [0,1] ]; let xs,ys; let testXs,testYs; let neuralNet; function parseData(){ let samples=rawData.split("\n"); shuffleArray(samples); xs=new Array(samples.length-120); ys=new Array(samples.length-120); testXs = new Array(120); testYs = new Array(120); for(let i=0;i 0; i--) { var j = Math.floor(Math.random() * (i + 1)); var temp = array[i]; array[i] = array[j]; array[j] = temp; } } function addPolynomials(varlist,degree){ polys=[]; for(let combo=0;combo<(degree)**varlist.length;combo++){ comboStr=combo.toString(degree).padStart(varlist.length,"0"); prod=1; for(let v=0;v{ h.history.loss.forEach((l)=>console.log("cost: "+l)); }); } function scale(xs){ let scaled=new Array(xs.length); for(let i=0;i=max?x:max; min=x<=min?x:min; } range=max-min; ranges.push(max-min); mins.push(min); for(let i=0;i{ return sample.map((feature,j)=>{ //if(j!=0){ return (testXs[i][j]-mins[j]-ranges[j]/2);//}else{ // return 1; // } }) }); inputsTNSR=tf.tensor(xs); outputsTNSR=tf.tensor(ys); testInputsTNSR=tf.tensor(testXs); initNeuralNet(inputsTNSR,outputsTNSR) .then(()=>{ let preds= neuralNet.predict(testInputsTNSR).arraySync(); let accurates=0; preds.forEach((p,i)=>{ if(p.indexOf(Math.max(...p))==testYs[i].indexOf(1)) accurates++; }); console.log(`${accurates}/120 accuracy on test set. ${100*accurates/120}%`); let preds2= neuralNet.predict(inputsTNSR).arraySync(); accurates=0; preds2.forEach((p,i)=>{ if(p.indexOf(Math.max(...p))==ys[i].indexOf(1)) accurates++; }); console.log(`${accurates}/186 accuracy on training set. ${100*accurates/186}%`); }); }

haberman survival