Opportunity <\/h3>\n\n\n\n
In order to understand\npredictive modelling, we tried to build a machine\nlearning model to accurately predict whether or not the patients in the dataset\nhave diabetes or not. The datasets consist of several\nmedical predictor variables and one target variable, Outcome. Predictor\nvariables includes the number of pregnancies the patient has had, their BMI,\ninsulin level, age, and so on.<\/p>\n\n\n\n
- Pregnancies<\/strong>: Number of times pregnant<\/li>
- Glucose<\/strong>: Plasma glucose concentration 2 hours\nin an oral glucose tolerance test<\/li>
- blood pressure<\/strong>: Diastolic blood\npressure (mm Hg)<\/li>
- SkinThickness<\/strong>: Triceps skinfold thickness (mm)<\/li>
- Insulin<\/strong>: 2-Hour serum insulin (mu U\/ml)<\/li>
- BMI<\/strong>: Body mass index (weight in\nkg\/(height in m)^2)<\/li>
- DiabetesPedigreeFunction<\/strong>: Diabetes\npedigree function<\/li>
- Age<\/strong>: Age (years)<\/li>
- Outcome<\/strong>: Class variable (0 or 1)<\/li><\/ul>\n\n\n\n
First five rows of the dataset look like this –<\/strong><\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-49.png\")
<\/figure><\/div>\n\n\n\nDataFrame.describe()<\/strong> method generates descriptive statistics that summarize the central tendency, dispersion and shape of a dataset\u2019s distribution, excluding NaN values. This method tells us a lot of things about a dataset.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-50.png\")
<\/figure><\/div>\n\n\n\nA question which arises here is \u2013<\/p>\n\n\n\n
Can the minimum value of below-listed\ncolumns be zero (0)?<\/h4>\n\n\n\n
On these columns, a value of zero does\nnot make sense and thus indicates a missing value. Following columns or\nvariables have an invalid zero value:<\/p>\n\n\n\n
- Glucose<\/li>
- blood pressure<\/li>
- skin thickness<\/li>
- Insulin<\/li>
- BMI<\/li><\/ol>\n\n\n\n
So we can say that the zero elements might be the null values in the dataset. Let\u2019s find out the Total null values in the dataset with respect to each feature.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-51.png\")
<\/figure><\/div>\n\n\n\nWe need to\nfigure out how to impute these NaN values so that we can use them in modelling.\nThere are many ways to impute NaN values in the dataset, such as <\/p>\n\n\n\n
- Filling in median values of\nthe features<\/li>
- Filling in the mean value\nof the total feature<\/li>
- Removing the feature column\n(if more than 70% of feature is null)<\/li><\/ul>\n\n\n\n
Here, to find out which way to do, we can use hist<\/strong> from matplotlib<\/strong> to know how the data is spread for a particular feature. <\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-52.png\")
<\/figure><\/div>\n\n\n\nWe will be imputing NaN values for the columns in accordance with their distributions.\nGlucose feature will use mean of the column as it is normally distributed,\nInsulin is skewed towards left and does have more values from 0-100, so we will\nuse median on it, same for BMI and Skin thickness features and at last for blood pressure we will use mean\non it.<\/p>\n\n\n\n
The results are \u2013<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-53.png\")
<\/figure><\/div>\n\n\n\nLet us see our outcome<\/strong> feature (this is our target variable) \u2013 <\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-54.png\")
<\/figure><\/div>\n\n\n\nNow let us see the Correlation between each variable to analyse our data well, we are using a seaborn<\/strong> package to get this correlation.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-55.png\")
<\/figure><\/div>\n\n\n\nGlucose and Insulin are the important features which might have a huge impact on the target variable (Outcome).<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-56.png\")
<\/figure><\/div>\n\n\n\nStandardisation<\/strong>: this is an important step before modelling as the features are needed to be in a particular range. Consider this example, if feature age 50 and insulin 0.5 is fed into ML algorithm the machine thinks that age is more important as it has higher value. For this reason, there is a need to normalize the dataset leaving out the target column. After completing this standardisation process the dataset looks like this- <\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-57.png\")
<\/figure><\/div>\n\n\n\nPredictive modelling:<\/strong><\/p>\n\n\n\n
We will be using different machine learning algorithms, ensemble\nmodelling, cross-validation(K-fold) to find out better accuracy for predicting\nif a person has Diabetes or not.<\/p>\n\n\n\n
SVM: <\/strong>we will be using support vector machine from sklearn python package. There are two kernels in svm namely Linear and rbf, and we will be testing both the algorithms.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-58.png\")
<\/figure><\/div>\n\n\n\nLogistic Regression: <\/strong>this algorithm is also from sklearn python package, here the main hyperparameters are C and Penalty(L1 or L2)<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-59.png\")
<\/figure><\/div>\n\n\n\nKNN: <\/strong>K- nearest neighbour is a machine learning algorithm which uses distance metrics to find the closest neighbours of our features, we need to find the value of k to get the best accuracy at a particular value.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-60.png\")
<\/figure><\/div>\n\n\n\nRandom Forest<\/strong>: this algorithm is a type of ensemble model which works well for classification problems. Let us first find the important features and then predict the outcome. These are the feature importance of our dataset – <\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-61.png\")
<\/figure><\/div>\n\n\n\nHere we can see that Glucose is the most important feature in the dataset if the Glucose level increases or Decreases then there is a huge possibility of getting Diabetes. <\/p>\n\n\n\n
Accuracy: <\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-62.png\")
<\/figure><\/div>\n\n\n\nCross-validation<\/strong>:<\/strong> Cross-validation\nis a resampling procedure used to evaluate machine learning models on a limited\ndata sample.<\/p>\n\n\n\n
The general procedure is as follows:<\/p>\n\n\n\n
- Shuffle the dataset randomly.<\/li>
- Split the dataset into k groups.<\/li>
- For each unique group:
- Take the group as a holdout or test data set.<\/li><\/ol>
- Take the remaining groups as a training data set.<\/li><\/ol>
- Fit a model on the training set and evaluate it on the test set.<\/li><\/ol>
- Retain the evaluation score and discard the model.<\/li><\/ol><\/li>
- Summarize the skill of the model using the sample of model evaluation\nscores.<\/li><\/ol>\n\n\n\n
Now we will be using this cross validation for our algorithms and check which gives us better accuracy \u2013 let K be 10<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-63.png\")
<\/figure><\/div>\n\n\n\nWe can infer\nthat the best accuracy for this model was from k-fold(Logistic regression) with\n77.5%<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-64.png\")
<\/figure><\/div>\n\n\n\nEnsemble modelling: <\/strong>we will be using Voting classifier in ensemble modeling , the two algorithms which got high accuracy before will be used in this \u2013 Linear SVM and Logistic Regression.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-65.png\")
<\/figure>\n\n\n\nWhile using\nensemble model, we couldn\u2019t get much better accuracy than single classifier, so\nthis method will not be used further.<\/p>\n\n\n\n
Best model:<\/strong><\/p>\n\n\n\n
As we have tried all the algorithms and method that we have mentioned above , best accuracy is got from Logistic Regression using K-Fold cross-validation.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-66.png\")
<\/figure><\/div>\n\n\n\nTP<\/strong> : 439 ; TN<\/strong> : 150 ; FP<\/strong> : 61 ; FN<\/strong> : 118 ;<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-67.png\")
<\/figure><\/div>\n\n\n\nAUC \u2013 ROC curve: <\/strong><\/p>\n\n\n\n
AUC\u2013ROC curve is the model selection metric for bi\u2013multi-class classification problem. ROC is a probability curve for different classes. ROC tells us how good the model is for distinguishing the given classes, in terms of the predicted probability. A typical ROC curve has False Positive Rate (FPR) on the X-axis and True Positive Rate (TPR) on the Y-axis. The area covered by the curve is the area between the orange line (ROC) and the axis. This area covered is AUC. The bigger the area covered, the better the machine learning models is at distinguishing the given classes. Ideal value for AUC is 1.<\/p>\n\n\n\n
![\"\"](\"https:\/\/guires.uk\/newsroom\/wp-content\/uploads\/2019\/11\/image-68.png\")
<\/figure><\/div>\n\n\n\nFurther\nproceedings: <\/p>\n\n\n\n
- This dataset is small\nand did not capture features well. Therefore, there is a need for more data in\norder to train the model to gain rich insights on the features.
- We can add few more\nfeatures which can be useful in predictions, new features should be added with\nconsent of Doctors which actually affects the outcome accurately. <\/li><\/ul><\/li><\/ul>\n\n\n\n
Why Guires<\/h3>\n\n\n\n
Guires\nData analytics mission is to democratize AI for healthcare industries. The team\nof data science expert use the power of AI to solve business and social\nchallenges. We are a pioneer in the\nresearch field for more than fifteen years and offer end to end solution for\nthe firm to set the direction for the company and support analytical frameworks\nfor better understanding and making strategic decisions. We provide appropriate\nsolutions using your existing volume of data available in varying degree of\ncomplexities that cannot be processed using traditional technologies,\nprocessing methods, or any commercial off the shelf solutions. By outsourcing\nbig data to us, we can analyze events that have happened within and outside an\norganization and correlate those to provide near accurate insights into what\ndrove the outcome. Our big data analytics solutions are fast, scalable and\npossess flexible processing.<\/p>\n\n\n\n
We use powerful algorithms, business rules, and statistical models. We work with text, image, audio, video and\nmachine data. Our medical experts understand the different layers of data being\nintegrated and what granularity levels of integration can be completed to\ncreate the holistic picture. Our team creates the foundational structure for\nanalytics and visualization of the data. Our data analytics team is well\nequipped with advanced mathematical degrees, statisticians with multiple\nspecialist degrees who can apply cutting-edge data mining techniques thereby\nenabling our clients to gain rich insights into existing customers and unearth\nhigh potential prospects.<\/p>\n\n\n\n
How\ncan you make the most of predictive analytics? Let us help you get started.<\/p>\n\n\n\n
Get\npredictive analytics working for you. Contact Guires expert.<\/p>\n\n\n\n
How\ncan you make the most of predictive analytics? Let us help you get started.<\/p>\n\n\n\n
Get\npredictive analytics working for you. Contact Guires expert.<\/p>\n","protected":false},"excerpt":{"rendered":"
The Challenges Diabetes is a group of metabolic disorders in which there are high blood sugar levels over a prolonged period. Symptoms of high blood sugar include frequent urination, increased thirst, and increased hunger. If left untreated, diabetes can cause many complications. Acute complications can include diabetic ketoacidosis, hyperosmolar hyperglycemic state, or death. Serious long-term […]<\/p>\n","protected":false},"author":1,"featured_media":624,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[55],"tags":[],"class_list":["post-619","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-use-cases"],"_links":{"self":[{"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/posts\/619","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/comments?post=619"}],"version-history":[{"count":2,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/posts\/619\/revisions"}],"predecessor-version":[{"id":760,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/posts\/619\/revisions\/760"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/media\/624"}],"wp:attachment":[{"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/media?parent=619"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/categories?post=619"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/guires.uk\/newsroom\/wp-json\/wp\/v2\/tags?post=619"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}
- We can add few more\nfeatures which can be useful in predictions, new features should be added with\nconsent of Doctors which actually affects the outcome accurately. <\/li><\/ul><\/li><\/ul>\n\n\n\n
- This dataset is small\nand did not capture features well. Therefore, there is a need for more data in\norder to train the model to gain rich insights on the features.
- Fit a model on the training set and evaluate it on the test set.<\/li><\/ol>
- Take the remaining groups as a training data set.<\/li><\/ol>
- Take the group as a holdout or test data set.<\/li><\/ol>
- Glucose<\/strong>: Plasma glucose concentration 2 hours\nin an oral glucose tolerance test<\/li>