Peru is a country with a complex geography. Moreover, it lacks minimal safe roads, higher in Andean and jungle territories. Depending on the classification they might receive, the num-ber of kilometers without pavement can reach up to 80%. Such a rate is alarming, consider-ing that internationally the United Nations promotes actions for road security since road ac-cidents are among the most common causes of death causes in the world. Then, employing data from Sutran, the current research has analyzed how the severity of accidents can affect road administration in Peru roads in 2021. Hence, the multinomial logistic regression was employed due to the nature of the data. It was found that regional routes had a higher risk than national roads when estimating the probability of having many dead and injured people because of a road accident.

The Central Road is the most connected road between the capital and cities of the jungle and the highlands in Peru. Nowadays, this road suffers from intense traffic jams that excessively delay time travel. Moreover, it is vulnerable to closures due to natural and human factors. Also, there is a proliferation of reckless drivers who do not care about speeding and cause terrible accidents. Hence, the current research explores whether the speeding record affected the occurrence of accidents on this road. After applying the Logit regression, it was possible to find that only recording speeding beyond 30 kilometers per hour over the limit was relevant to prevent accidents. It was also suggested that the speedometer should track 24 hours a day along with measures to improve the infrastructure of this road.

Highway accident data analysis provides probability of occurrence of road accidents by associating different accident factors using data mining algorithms. Analysis can be improved by using advanced data mining algorithms that compute relationships with minimum processing time. As accident datasets are very heterogeneous in nature, it is difficult to identify the relationship between critical factors responsible for road accidents without data mining algorithms. In this study, K-modes for clustering and frequent pattern growth algorithms to extract relationships between critical accident factors have been used. The accomplished result concludes better relationships with better accuracy than earlier implemented data mining algorithms and has found meaningful hidden situations that would be beneficial for future work in decreasing the number of highway accidents.