Eliminating Last-Mile Inefficiencies in the Trucking Industry
Pilot Freight Services, traditionally a bulk cargo freight forwarder in the US, is in the process of expanding their business to provide last-mile delivery (LMD) services. This capstone project helps Pilot improve the performance of their LMD operations through higher visibility and elimination of e...
| Main Authors: | , |
|---|---|
| Format: | Other |
| Language: | en_US |
| Published: |
2020
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/1721.1/126494 |
| _version_ | 1811073015577837568 |
|---|---|
| author | From, Kristian Mangan, Katharina |
| author_facet | From, Kristian Mangan, Katharina |
| author_sort | From, Kristian |
| collection | MIT |
| description | Pilot Freight Services, traditionally a bulk cargo freight forwarder in the US, is in the process of expanding their business to provide last-mile delivery (LMD) services. This capstone project helps Pilot improve the performance of their LMD operations through higher visibility and elimination of efficiencies. First, an understanding of Pilot’s current LMD operation is established. Next, a performance metric framework is defined, with two performance dimensions: (1) service level and (2) efficiency. Guided by the framework, the performance of Pilot’s LMD operations is assessed by analyzing descriptive statistics. A visualization tool is built in Tableau, allowing Pilot to continuously assess their own performance. Finally, machine learning is used to identify parameters that affect performance and predict their impact. The parameters identified as having the biggest impact on stop time duration are: volume delivered, population density, quantity pieces delivered, stop number, time of day, and peak day. For drive time duration, the single most relevant factor is mileage. For each of the locations analyzed, coefficients are calculated and made available to Pilot’s planners to predict stop and drive time based on the parameters. Planning accuracy, in terms of MAPE, is for stop time improved from about 85% to about 55%, and for drive time from about 45% to 25%. The insight provided by this capstone will allow Pilot to better understand and assess the performance of their LMD operations and help identify areas for improvement. |
| first_indexed | 2024-09-23T09:27:19Z |
| format | Other |
| id | mit-1721.1/126494 |
| institution | Massachusetts Institute of Technology |
| language | en_US |
| last_indexed | 2024-09-23T09:27:19Z |
| publishDate | 2020 |
| record_format | dspace |
| spelling | mit-1721.1/1264942020-08-07T04:22:31Z Eliminating Last-Mile Inefficiencies in the Trucking Industry From, Kristian Mangan, Katharina machine learning data analytics transportation Pilot Freight Services, traditionally a bulk cargo freight forwarder in the US, is in the process of expanding their business to provide last-mile delivery (LMD) services. This capstone project helps Pilot improve the performance of their LMD operations through higher visibility and elimination of efficiencies. First, an understanding of Pilot’s current LMD operation is established. Next, a performance metric framework is defined, with two performance dimensions: (1) service level and (2) efficiency. Guided by the framework, the performance of Pilot’s LMD operations is assessed by analyzing descriptive statistics. A visualization tool is built in Tableau, allowing Pilot to continuously assess their own performance. Finally, machine learning is used to identify parameters that affect performance and predict their impact. The parameters identified as having the biggest impact on stop time duration are: volume delivered, population density, quantity pieces delivered, stop number, time of day, and peak day. For drive time duration, the single most relevant factor is mileage. For each of the locations analyzed, coefficients are calculated and made available to Pilot’s planners to predict stop and drive time based on the parameters. Planning accuracy, in terms of MAPE, is for stop time improved from about 85% to about 55%, and for drive time from about 45% to 25%. The insight provided by this capstone will allow Pilot to better understand and assess the performance of their LMD operations and help identify areas for improvement. Pilot Freight Services 2020-08-06T19:43:42Z 2020-08-06T19:43:42Z 2020-08-06 Other https://hdl.handle.net/1721.1/126494 en_US application/pdf |
| spellingShingle | machine learning data analytics transportation From, Kristian Mangan, Katharina Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title | Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title_full | Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title_fullStr | Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title_full_unstemmed | Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title_short | Eliminating Last-Mile Inefficiencies in the Trucking Industry |
| title_sort | eliminating last mile inefficiencies in the trucking industry |
| topic | machine learning data analytics transportation |
| url | https://hdl.handle.net/1721.1/126494 |
| work_keys_str_mv | AT fromkristian eliminatinglastmileinefficienciesinthetruckingindustry AT mangankatharina eliminatinglastmileinefficienciesinthetruckingindustry |