Thank you for being up to the challenge. We are proud that you have tried and did your best. Your choice of data visualization was interesting, considering this is a regression based model. Every variable in the model is a partial effect to the overdose deaths. You are correct when your report states that there are things beyond opioid potency that we need to consider. If we do more scatter plots, we see that potency is a very weak positive relationship. Furthermore, there is the clustering effect. This should have been written in the report. Log transformation helps visualize the extreme potency differences. What shows a clearer relationship however, is the relationship between deaths and treatment accessibility. Treatment access in this case means access to individualized treatment options. The negative coefficient (-2.7268) suggests that 1% increase in accessibility associated with ~2.73% decrease in death rates. Marginally significant (p=0.098). This means, stronger effect than potency (ẞ1=0.0088). A part of this project is to refine the model: How can the model be improved? What have we gathered from the data ranging in 1980s to 2020? We see that the overdose deaths were somewhat linear back when the supply was mainly natural opioids. As the potency goes up, the curve shows an exponential growth. This is where the model needs to reflect the reality of this exponential curve . For synthetic opioids: Death Rate = α * e^(β*potency) For non-synthetic opioids: Death Rate = α + linearFactor * potency Where: α = base death rate β = exponential growth factor linearFactor = linear growth rate for non-synthetic opioids. I hope you learned a lot from this project, thank you so much for working with us!