A linear regression machine learning model derives the linear relationship between single or multiple features (put in the x-axis of a co-ordinate plane) and a single response (put in the y-axis of the co-ordinate plane) for a given set of observations. The model then learns to predict the response for a set of new feature values using the derived relationship. However, the linear regression model does not have the flexibility to predict the feature values for a target response. The solution proposed in this paper can leverage the relationship derived by the linear regression model between multiple features and single response. Using the relationship, it can predict the feature values for a target response value. In the proposed solution, the model accepts the training data in two separate input datasets – one contains the features in observations and the other contains the responses. After making the prediction on feature values for a queried response value, the model returns a two dimensional array of numbers. Each column of the output array contains the predicted values for a specific feature. Each row of the array contains different valid sets of feature values. Each set of feature values results the queried response value according to their linear relationship.