Methods and systems for selecting a machine learning algorithm
Methods and systems for selecting a machine learning algorithm are described. In one embodiment, one or more factors to be used by a machine learning algorithm in predicting a value of a required pharmacy element of a prescription are identified, the machine learning algorithm is trained to predict the value of the required pharmacy element using a first subset of previously received prescriptions, a success rates for the machine learning algorithm at predicting respective known values of respective known required pharmacy elements for each of a second subset of the previously received prescriptions are determined, and the machine learning algorithm predicts the value of the required pharmacy element of the prescription for a first predetermined period.
1. A method comprising:
identifying, by a machine learning algorithm selection subsystem implemented by a processor, one or more factors to be used by a machine learning algorithm in predicting a value of an element entered into a form;
training, by the machine learning algorithm selection subsystem, the machine learning algorithm to predict the value of the element using a first subset of previously received data by analyzing known element values within the first subset of previously received data and a relationship between the known element values and the one or more factors;
predicting, by the machine learning algorithm, respective known values of respective known elements for each of a second subset of previously received data, the second subset comprising a remainder of the previously received data not included within the first subset;
evaluating, by the machine learning algorithm selection subsystem, a success rate for the machine learning algorithm at predicting respective known values of respective known elements for each of the second subset by determining whether the machine learning algorithm correctly predicted the respective known values of respective known elements for each of the second subset;
receiving, by the machine learning algorithm, a new form subsequent to training and evaluating the machine learning algorithm;
predicting, by the machine learning algorithm, the value of the element in the new form; and
pre-populating, by the machine learning algorithm, the value of the element in the new form, as predicted.
2. The method of claim 1 , wherein the machine learning algorithm comprises K-Neighbor, Random Forest, Gaussian Naive Bayes, and Stochastic gradient descent.
3. The method of claim 1 , wherein the first subset contains 80% of the previously received data.
4. The method of claim 3 , wherein the second subset contains the remaining 20% of the previously received data not included in the first subset.
5. The method of claim 1 , further comprising:
a prediction subsystem determining a confidence value for predicting the value of the element in the new form.
6. The method of claim 5 , further comprising:
the prediction subsystem determining whether the confidence value exceeds a threshold,
wherein the machine learning algorithm pre-populates the value of the element in the new form with the value of the element, as predicted, when the confidence value exceeds the threshold.
7. The method of claim 6 , wherein the threshold is 80%.
8. A system comprising:
a storage device to store a machine learning algorithm;
a subsystem in communication with the storage device and configured to:
identify one or more factors to be used by a machine learning algorithm in predicting a value of an element entered into a form;
train the machine learning algorithm to predict the value of the element using a first subset of previously received data by analyzing known element values within the first subset of previously received data and a relationship between the known element values and the one or more factors;
predict respective known values of respective known elements for each of a second subset of previously received data, the second subset comprising a remainder of the previously received data not included within the first subset;
evaluate a success rate for the machine learning algorithm at predicting respective known values of respective known elements for each of the second subset by determining whether the machine learning algorithm correctly predicted the respective known values of respective known elements for each of the second subset;
a prediction subsystem configured to:
receive a new form subsequent to training and evaluating the machine learning algorithm;
predict the value of the element in the new form using the machine learning algorithm; and
pre-populate the value of the element in the new form, as predicted, by the machine learning algorithm.
9. The system of claim 8 , wherein the machine learning algorithm comprises K-Neighbor, Random Forest, Gaussian Naive Bayes, and Stochastic gradient descent.
10. The system of claim 8 , wherein the first subset contains 80% of the previously received data.
11. The system of claim 10 , wherein the second subset contains a remaining 20% of the previously received data not included in the first subset.
12. The system of claim 8 , wherein the subsystem at least includes a processor.
13. The system of claim 8 , wherein the prediction subsystem is configured to determine a confidence value for predicting the value of the element in the new form.
14. The system of claim 13 , wherein the prediction subsystem is further configured to determine whether the confidence value exceeds a threshold, wherein the prediction subsystem pre-populates the value of the element in the new form with the value of the element, as predicted, when the confidence value exceeds the threshold.
15. The system of claim 14 , wherein the threshold is 80%.
16. A non-transitory machine-readable medium comprising instructions, which, when executed by one or more processors, cause the one or more processors to perform the following operations:
identify one or more factors to be used by a machine learning algorithm in predicting a value of an element entered into a form;
train the machine learning algorithm to predict the value of the element using a first subset of previously received data by analyzing known element values within the first subset of previously received data and a relationship between the known element values and the one or more factors;
predict respective known values of respective known elements for each of a second subset of previously received data, the second subset comprising a remainder of the previously received data not included within the first subset;
evaluate a success rate for the machine learning algorithm at predicting respective known values of respective known elements for each of the second subset by determining whether each of the machine learning algorithm correctly predicted the respective known values of respective known elements for each of the second subset;
receive a new form subsequent to training and evaluating the machine learning algorithm;
predict the value of the element in the new form using the machine learning algorithm; and
pre-populate the value of the element in the new form, as predicted, by the machine learning algorithm.
17. The non-transitory machine-readable medium of claim 16 , wherein non-transitory machine-readable medium comprising instructions, which, when executed by the one or more processors, further cause the one or more processors to perform the following operations:
determine a confidence value for predicting the value of the element in the new form.
18. The non-transitory machine-readable medium of claim 17 , wherein non-transitory machine-readable medium comprising instructions, which, when executed by the one or more processors, further cause the one or more processors to perform the following operations:
determine whether the confidence value exceeds a threshold, wherein the one or more processors pre-populates the value of the element in the new form with the value of the element, as predicted, when the confidence value exceeds the threshold.
19. The non-transitory machine-readable medium of claim 18 , wherein the threshold is 80%.
20. The non-transitory machine-readable medium of claim 16 , wherein the machine learning algorithm comprises at least one of K-Neighbor, Random Forest, Gaussian Naive Bayes, Stochastic gradient descent, or combinations thereof, and wherein the machine learning algorithm determines a confidence value for predicting the value of the element.