• Nenhum resultado encontrado

As mentioned in Deployment section, the final solution of this project is a dashboard built with Power BI being divided in two parts, an Initial Analysis and a Cluster Analysis. Looking to Initial Analysis pages in Figure 31, initially there were 379 different locations, there were a total of 486 indicators, however a significant part of the locations only had 183 different indicators. Similarly, the majority of the locations had 9 different subthemes from a total of 17. Almost 90% of the locations belonged to renewables activity and around 75% percent of all the locations were in wind technology. Few locations were new and even less had contributors’ changes by comparing with the previous filling period, this is comprehensible as new locations’ contributors were not considered as having changed, otherwise these values would not be possible. Most of the locations followed what was recommended by our team to become the process more efficient and inserted the values through Excel files.

Regarding the distribution of number of indicators for each subtheme, it is evident that there is an unbalanced distribution, being that in each subtheme there is always a value with 275 or more locations. Important to underline that waste is the subtheme with the highest number of indicators associated to a location, with some locations having more than 200 indicators of this subtheme, illustrating the difficulties and challenges faced with waste indicators in previous data collection periods.

Figure 32. Initial Analysis II and III Pages Figure 31. Initial Analysis I Page

41 Considering variables related with filling process, locations tend to upload few indicators simultaneously as more than 50% of them uploaded a maximum of 8 indicators simultaneously.

Looking to indicators uploaded late, only 28 locations fulfilled with the established deadline, furthermore, most of the locations did not comply with the deadline and had 18 or 19 late indicators.

Moreover, locations had a tendency of uploading various files with most of them uploading 4 different files. This lack of efficiency during the filling period, with few indicators uploaded simultaneously and several files used to upload the data, can justify the delays in the process and the high quantity of indicators uploaded late. Considering the doubts posed, most of the contributors had at least one doubt to clarify, tending to use for a higher number of times email instead of Microsoft Teams.

Certainly that most of the locations inserting the first values in the last day of the filling period did not contribute to comply with the deadline, namely if some doubts emerged, delaying the process and leading that around 93% of the locations have inserted the last values after the deadline. Considering the number of indicators uploaded by week, in first and second week the number is not significant, being in third week that most of the indicators were uploaded and having around 10% of the indicators uploaded after the deadline, being this value specially alarming.

Moving to the main focus of the project, the clustering of the locations and the characterization of each one (based on Figure 35 with Cluster Analysis I and II pages). Besides Cluster 1 being the cluster

Figure 34. Initial Analysis V Page Figure 33. Initial Analysis IV Page

42 with less locations, stands out for having more distinct themes while having less distinct subthemes and indicators. These locations also have more fleet and environmental management system indicators and less energy consumption, environmental complaints, emissions and other emissions indicators. They asked more for clarifications via Microsoft Teams and less via Email when comparing with the remaining clusters, also seemed to meet the deadline as they had less late indicators and uploaded less zero values. Regarding binary variables, Cluster 1 have a high proportion of locations with replaced values, has the lowest proportion of insertion via Excel, besides having the highest proportion of locations with changes in contributors do not show new locations, having any location using wind technology nor in renewables activity.

Cluster 2 has the locations that uploaded the highest number of indicators at once, with more indicators from other emissions, liquid effluents and energy consumption subthemes. Additionally, asked less for clarifications via Email than the remaining clusters and more via Microsoft Teams than clusters 3 and 4, also had less late indicators, besides having more indicators in the total. The locations in this cluster show the highest average of zero values uploaded and distinct unit measures. In this cluster there are some locations with changes in contributors and replaced and validated values, all the locations inserted the values with Excel, there are not new locations neither use wind technology nor in renewable activity.

Regarding cluster 3, it shows the lowest average number of late indicators, less days between first and last insertion, having less doubts via Email than clusters 1 and 4 and less doubts via Microsoft Teams than clusters 1 and 2. These locations have less distinct subthemes, distinct unit measures, total number of indicators and zero values inserted. In addition, they have less emissions and environmental complaints indicators and a high number of environmental management system indicators. This is the cluster with highest proportion locations with validated values not having replaced values, some of the locations did not upload the values via Excel, has the highest proportion of new locations with 40% of the locations being in renewables activity.

Cluster 4 contains the highest number of locations, a total of 317 to be precise, having more late indicators and doubts via Email than the remaining clusters. These locations also took more time between the first and last values uploaded, having less distinct themes, new contributors and environmental management system indicators. In this cluster, a very small proportion of the locations replaced and validated the values inserted, some of the locations are new, is the only cluster without changes in the contributors and with some not consolidated locations. One significant difference between this cluster and the remaining clusters is that around 90% of the locations use wind technology and almost 100% are in renewables activity.

Looking more into detal to Cluster Analysis I and II pages in Figure 35 and comparing with the description of each cluster above, some overall conclusions can be considered:

• Locations with less indicators tend to have less late indicators, namely in clusters 1 and 3;

• Locations with more new contributors tend to have preference for asking for clarifications via Microsoft Teams than via Email, namely in clusters 1 and 2;

• Locations with more late indicators and less new contributors tend to ask more through Email, namely in cluster 4;

• Locations with less distinct subthemes and indicators uploaded less zero values and took less days to upload all the data, namely in clusters 1 and 3.

43 Figure 35. Cluster Analysis I and II Pages

44

Documentos relacionados