Bidders Recommender for Public Procurement Auctions Using Machine Learning: Data Analysis, Algorithm, and Case Study with Tenders from Spain

Table 5

Testing the bidders recommender for two scenarios: results of the accuracy and number of recommended companies per tender for five different setups.

Description		Different bidders recommender settings
		Very low	Low	Medium	High	Very high
Bidders recommender factors for the settings	: operating income factor	0.25	0.5	0.65	0.75	1.0
	: EBIT factor	0.25	0.5	0.65	0.75	1.0
	: EBITDA factor	0.25	0.5	0.65	0.75	1.0
	: employees factor	0.15	0.25	0.25	0.35	0.45
	: classification economic activities factor	0.125	0.15	0.14	0.175	0.2
	: distance tender-company factor	1.6	1.4	1.4	1.2	1
Results of scenario 1: testing subset is the 20% of the dataset randomly chosen	: winner company is the forecast company	17.07%
	: winner company is within the top 5 forecast companies	31.58%
	: winner company is within the recommended companies group	38.52%	36.20%	35.92%	34.04%	33.25%
	Mean and median number of the recommended companies of each tender	877.43; 86	469.69; 35	430.48; 31	226.07; 11	145.97; 9
Results of scenario 2: testing subset is the last 20% of the dataset ordered by tender’s date	: winner company is the forecast company	10.25%
	: winner company is within the top 5 forecast companies	23.12%
	: winner company is within the recommended companies group	30.52%	28.00%	27.73%	25.55%	24.79%
	Mean and median number of the recommended companies of each tender	900.64; 95	470.41; 37	430.33; 33	210.92; 11	132.10; 9