# ks_tasks
The folder "ks_tasks" contains the generated training data for our models and the testset "ks_tasks".
"ks_tasks" contains the knapsack instances that have been shown to the participants of our user study.
The recommendation values are the values generated by the ML models. 

### Ignore q7,q8,q9,q10 those were dummy values generated with dynamic programming
### The naming in ks_tasks is different from the naming in the paper as follows:

	q1 --> q3 in the paper
	q2 --> q4 "
	q3 --> q6 "
	q4 --> q1 "
	q5 --> q2 "
	q6 --> q5 "


# models
The folder "models" contains the models used to produce the recommendations q1,...,q6 . The naming is correct as in the paper
with q1 being the worst model and q6 the best.

# model training
The folder "model training" contains the code to generate test and training data, train and evaluate models and 
store the models in the format required to display it in the webapp. The jupyter notebook contains additional
comments to navigate the reader through the code.
The file "knapsack_utils" contains all the code required to generate and solve the knapsack problem. It also contains
the test and training loops as well as the neural net architecture of our model.

# plots and analysis
The Plots and analysis folder contains our data cleaning and plot generation code.(Python)
It also contains the code for plotting and statistical analysis in R
The folder contains the anonymized and cleaned data that we collected in our user study and the raw data.