Machine Learning

how to start to teach the machine

@ArkadiuszKondas

ARKADIUSZ KONDAS

Lead Software Architect
@ Proget Sp. z o.o.
Poland

Zend Certified Engineer

Code Craftsman

Blogger

Ultra Runner

@ ArkadiuszKondas

arkadiuszkondas.com

Zend Certified Architect

Agenda:

  1. Why Machine Learning?
  2. Ways of learning
  3. How to predict realty price in Russia?
  4. Summary

Why machine learning?

Why Machine Learning?

  • Develop systems that can automatically adapt and customize themselves to individual users.
    • personalized news or mail filter

Why Machine Learning?

  • Discover new knowledge from large databases (data mining).
    • market basket analysis

Source: https://blogs.adobe.com/digitalmarketing/analytics/shopping-for-kpis-market-basket-analysis-for-web-analytics-data/

Why Machine Learning?

  • Ability to mimic human and replace certain monotonous tasks - which require some intelligence.
    • like recognizing handwritten characters

https://github.com/tensorflow/models/tree/master/im2txt

Why Machine Learning?

http://articles.concreteinteractive.com/nicole-kidmans-fake-nose/

Why Machine Learning?

  • Develop systems that are too difficult/expensive to construct manually because they require specific detailed skills or knowledge tuned to a specific task (knowledge engineering bottleneck)

Why now?

The name machine learning was coined in 1959

  • Data availability
     
  • Computation power

Ways of learning

Supervised learning

Source: https://www.slideshare.net/Simplilearn/what-is-machine-learning-machine-learning-basics-machine-learning-algorithms-simplilearn

Supervised learning

Unsupervised learning

Source: https://www.slideshare.net/Simplilearn/what-is-machine-learning-machine-learning-basics-machine-learning-algorithms-simplilearn

Unsupervised learning

Reinforcement learning

Source: https://www.marutitech.com/businesses-reinforcement-learning/

Reinforcement learning

How to predict realty price in Russia?

Supervised learning

Source: https://www.slideshare.net/Simplilearn/what-is-machine-learning-machine-learning-basics-machine-learning-algorithms-simplilearn

Sberbank Russian Housing Market

Can you predict realty price fluctuations in Russia’s volatile economy?

Dataset

id,timestamp,full_sq,life_sq,floor,max_floor,material,build_year,num_room,kitch_sq,state,product_type,sub_area,area_m,raion_popul,green_zone_part,indust_part,children_preschool,preschool_quota,preschool_education_centers_raion,children_school,school_quota,school_education_centers_raion,school_education_centers_top_20_raion,hospital_beds_raion,healthcare_centers_raion,university_top_20_raion,sport_objects_raion,additional_education_raion,culture_objects_top_25,culture_objects_top_25_raion,shopping_centers_raion,office_raion,thermal_power_plant_raion,incineration_raion,oil_chemistry_raion,radiation_raion,railroad_terminal_raion,big_market_raion,nuclear_reactor_raion,detention_facility_raion,full_all,male_f,female_f,young_all,young_male,young_female,work_all,work_male,work_female,ekder_all,ekder_male,ekder_female,0_6_all,0_6_male,0_6_female,7_14_all,7_14_male,7_14_female,0_17_all,0_17_male,0_17_female,16_29_all,16_29_male,16_29_female,0_13_all,0_13_male,0_13_female,raion_build_count_with_material_info,build_count_block,build_count_wood,build_count_frame,build_count_brick,build_count_monolith,build_count_panel,build_count_foam,build_count_slag,build_count_mix,raion_build_count_with_builddate_info,build_count_before_1920,build_count_1921-1945,build_count_1946-1970,build_count_1971-1995,build_count_after_1995,ID_metro,metro_min_avto,metro_km_avto,metro_min_walk,metro_km_walk,kindergarten_km,school_km,park_km,green_zone_km,industrial_km,water_treatment_km,cemetery_km,incineration_km,railroad_station_walk_km,railroad_station_walk_min,ID_railroad_station_walk,railroad_station_avto_km,railroad_station_avto_min,ID_railroad_station_avto,public_transport_station_km,public_transport_station_min_walk,water_km,water_1line,mkad_km,ttk_km,sadovoe_km,bulvar_ring_km,kremlin_km,big_road1_km,ID_big_road1,big_road1_1line,big_road2_km,ID_big_road2,railroad_km,railroad_1line,zd_vokzaly_avto_km,ID_railroad_terminal,bus_terminal_avto_km,ID_bus_terminal,oil_chemistry_km,nuclear_reactor_km,radiation_km,power_transmission_line_km,thermal_power_plant_km,ts_km,big_market_km,market_shop_km,fitness_km,swim_pool_km,ice_rink_km,stadium_km,basketball_km,hospice_morgue_km,detention_facility_km,public_healthcare_km,university_km,workplaces_km,shopping_centers_km,office_km,additional_education_km,preschool_km,big_church_km,church_synagogue_km,mosque_km,theater_km,museum_km,exhibition_km,catering_km,ecology,green_part_500,prom_part_500,office_count_500,office_sqm_500,trc_count_500,trc_sqm_500,cafe_count_500,cafe_sum_500_min_price_avg,cafe_sum_500_max_price_avg,cafe_avg_price_500,cafe_count_500_na_price,cafe_count_500_price_500,cafe_count_500_price_1000,cafe_count_500_price_1500,cafe_count_500_price_2500,cafe_count_500_price_4000,cafe_count_500_price_high,big_church_count_500,church_count_500,mosque_count_500,leisure_count_500,sport_count_500,market_count_500,green_part_1000,prom_part_1000,office_count_1000,office_sqm_1000,trc_count_1000,trc_sqm_1000,cafe_count_1000,cafe_sum_1000_min_price_avg,cafe_sum_1000_max_price_avg,cafe_avg_price_1000,cafe_count_1000_na_price,cafe_count_1000_price_500,cafe_count_1000_price_1000,cafe_count_1000_price_1500,cafe_count_1000_price_2500,cafe_count_1000_price_4000,cafe_count_1000_price_high,big_church_count_1000,church_count_1000,mosque_count_1000,leisure_count_1000,sport_count_1000,market_count_1000,green_part_1500,prom_part_1500,office_count_1500,office_sqm_1500,trc_count_1500,trc_sqm_1500,cafe_count_1500,cafe_sum_1500_min_price_avg,cafe_sum_1500_max_price_avg,cafe_avg_price_1500,cafe_count_1500_na_price,cafe_count_1500_price_500,cafe_count_1500_price_1000,cafe_count_1500_price_1500,cafe_count_1500_price_2500,cafe_count_1500_price_4000,cafe_count_1500_price_high,big_church_count_1500,church_count_1500,mosque_count_1500,leisure_count_1500,sport_count_1500,market_count_1500,green_part_2000,prom_part_2000,office_count_2000,office_sqm_2000,trc_count_2000,trc_sqm_2000,cafe_count_2000,cafe_sum_2000_min_price_avg,cafe_sum_2000_max_price_avg,cafe_avg_price_2000,cafe_count_2000_na_price,cafe_count_2000_price_500,cafe_count_2000_price_1000,cafe_count_2000_price_1500,cafe_count_2000_price_2500,cafe_count_2000_price_4000,cafe_count_2000_price_high,big_church_count_2000,church_count_2000,mosque_count_2000,leisure_count_2000,sport_count_2000,market_count_2000,green_part_3000,prom_part_3000,office_count_3000,office_sqm_3000,trc_count_3000,trc_sqm_3000,cafe_count_3000,cafe_sum_3000_min_price_avg,cafe_sum_3000_max_price_avg,cafe_avg_price_3000,cafe_count_3000_na_price,cafe_count_3000_price_500,cafe_count_3000_price_1000,cafe_count_3000_price_1500,cafe_count_3000_price_2500,cafe_count_3000_price_4000,cafe_count_3000_price_high,big_church_count_3000,church_count_3000,mosque_count_3000,leisure_count_3000,sport_count_3000,market_count_3000,green_part_5000,prom_part_5000,office_count_5000,office_sqm_5000,trc_count_5000,trc_sqm_5000,cafe_count_5000,cafe_sum_5000_min_price_avg,cafe_sum_5000_max_price_avg,cafe_avg_price_5000,cafe_count_5000_na_price,cafe_count_5000_price_500,cafe_count_5000_price_1000,cafe_count_5000_price_1500,cafe_count_5000_price_2500,cafe_count_5000_price_4000,cafe_count_5000_price_high,big_church_count_5000,church_count_5000,mosque_count_5000,leisure_count_5000,sport_count_5000,market_count_5000,price_doc
1,2011-08-20,43,27,4,NA,NA,NA,NA,NA,NA,Investment,Bibirevo,6407578.1,155572,0.189727117,6.99893e-5,9576,5001,5,10309,11065,5,0,240,1,0,7,3,no,0,16,1,no,no,no,no,no,no,no,no,86206,40477,45729,21154,11007,10147,98207,52277,45930,36211,10580,25631,9576,4899,4677,10309,5463,4846,23603,12286,11317,17508,9425,8083,18654,9709,8945,211,25,0,0,0,2,184,0,0,0,211,0,0,0,206,5,1,2.590241095,1.131259906,13.57511887,1.131259906,0.145699552,0.17797535,2.158587074,0.600973099,1.080934313,23.68346,1.804127,3.633334,5.419893032,65.03871639,1,5.419893032,6.905892968,1,0.274985143,3.299821714,0.992631058,no,1.42239141,10.9185867,13.10061764,13.67565705,15.15621058,1.422391404,1,no,3.830951404,5,1.305159492,no,14.23196091,101,24.2924061,1,18.152338,5.718518835,1.210027392,1.062513046,5.814134663,4.308127002,10.81417151,1.676258313,0.485841388,3.065047099,1.107594209,8.148590774,3.516512911,2.392353035,4.248035887,0.974742843,6.715025787,0.884350021,0.648487637,0.637188832,0.947961657,0.17797535,0.625783434,0.628186549,3.932040333,14.05304655,7.389497904,7.023704919,0.516838085,good,0,0,0,0,0,0,0,NA,NA,NA,0,0,0,0,0,0,0,0,0,0,0,1,0,7.36,0,1,30500,3,55600,19,527.78,888.89,708.33,1,10,4,3,1,0,0,1,2,0,0,6,1,14.27,6.92,3,39554,9,171420,34,566.67,969.7,768.18,1,14,11,6,2,0,0,1,2,0,0,7,1,11.77,15.97,9,188854,19,1244891,36,614.29,1042.86,828.57,1,15,11,6,2,1,0,1,2,0,0,10,1,11.98,13.55,12,251554,23,1419204,68,639.68,1079.37,859.52,5,21,22,16,3,1,0,2,4,0,0,21,1,13.09,13.31,29,807385,52,4036616,152,708.57,1185.71,947.14,12,39,48,40,9,4,0,13,22,1,0,52,4,5850000
2,2011-08-23,34,19,3,NA,NA,NA,NA,NA,NA,Investment,Nagatinskij Zaton,9589336.912,115352,0.372602044,0.049637257,6880,3119,5,7759,6237,8,0,229,1,0,6,1,yes,1,3,0,no,no,no,no,no,no,no,no,76284,34200,42084,15727,7925,7802,70194,35622,34572,29431,9266,20165,6880,3466,3414,7759,3909,3850,17700,8998,8702,15164,7571,7593,13729,6929,6800,245,83,1,0,67,4,90,0,0,0,244,1,1,143,84,15,2,0.936699728,0.647336757,7.620630408,0.635052534,0.147754269,0.273345319,0.550689737,0.065321162,0.966479097,1.317476,4.655004,8.648587,3.411993084,40.943917,2,3.641772591,4.679744508,2,0.065263344,0.78316013,0.698081318,no,9.503405157,3.103995954,6.444333466,8.132640073,8.698054189,2.887376585,2,no,3.103995974,4,0.694535727,no,9.242585522,32,5.706113234,2,9.034641872,3.489954443,2.72429538,1.246148739,3.419574049,0.725560431,6.910567711,3.424716092,0.668363679,2.000153804,8.97282283,6.127072782,1.161578983,2.543746975,12.64987875,1.47772267,1.852560245,0.686251693,0.519311324,0.688796317,1.072315063,0.273345319,0.967820571,0.471446524,4.841543888,6.829888847,0.709260033,2.358840498,0.23028691,excellent,25.14,0,0,0,0,0,5,860,1500,1180,0,1,3,0,0,1,0,0,1,0,0,0,0,26.66,0.07,2,86600,5,94065,13,615.38,1076.92,846.15,0,5,6,1,0,1,0,1,2,0,4,2,0,21.53,7.71,3,102910,7,127065,17,694.12,1205.88,950,0,6,7,1,2,1,0,1,5,0,4,9,0,22.37,19.25,4,165510,8,179065,21,695.24,1190.48,942.86,0,7,8,3,2,1,0,1,5,0,4,11,0,18.07,27.32,12,821986,14,491565,30,631.03,1086.21,858.62,1,11,11,4,2,1,0,1,7,0,6,19,1,10.26,27.47,66,2690465,40,2034942,177,673.81,1148.81,911.31,9,49,65,36,15,3,0,15,29,1,10,66,14,6000000
3,2011-08-27,43,29,2,NA,NA,NA,NA,NA,NA,Investment,Tekstil'shhiki,4808269.831,101708,0.112559644,0.118537385,5879,1463,4,6207,5580,7,0,1183,1,0,5,1,no,0,0,1,no,no,no,yes,no,no,no,no,101982,46076,55906,13028,6835,6193,63388,31813,31575,25292,7609,17683,5879,3095,2784,6207,3269,2938,14884,7821,7063,19401,9045,10356,11252,5916,5336,330,59,0,0,206,4,60,0,1,0,330,1,0,246,63,20,3,2.120998901,1.637996285,17.3515154,1.445959617,0.049101536,0.158071895,0.374847751,0.453172405,0.939275144,4.91266,3.381083,11.99648,1.277658039,15.33189647,3,1.277658039,1.701419537,3,0.328756044,3.945072522,0.468264622,no,5.60479992,2.927487097,6.963402995,8.054252314,9.067884956,0.647249803,3,no,2.927487099,4,0.70069112,no,9.540544478,5,6.710302485,3,5.777393501,7.50661249,0.772216104,1.60218297,3.682454651,3.562187704,5.75236835,1.375442778,0.733101062,1.239303854,1.978517187,0.767568769,1.952770629,0.621357002,7.682302975,0.097143527,0.841254102,1.510088854,1.48653302,1.543048836,0.391957389,0.158071895,3.178751487,0.755946015,7.92215157,4.273200485,3.156422843,4.958214283,0.190461977,poor,1.67,0,0,0,0,0,3,666.67,1166.67,916.67,0,0,2,1,0,0,0,0,0,0,0,0,0,4.99,0.29,0,0,0,0,9,642.86,1142.86,892.86,2,0,5,2,0,0,0,0,1,0,0,5,3,9.92,6.73,0,0,1,2600,14,516.67,916.67,716.67,2,4,6,2,0,0,0,0,4,0,0,6,5,12.99,12.75,4,100200,7,52550,24,563.64,977.27,770.45,2,8,9,4,1,0,0,0,4,0,0,8,5,12.14,26.46,8,110856,7,52550,41,697.44,1192.31,944.87,2,9,17,9,3,1,0,0,11,0,0,20,6,13.69,21.58,43,1478160,35,1572990,122,702.68,1196.43,949.55,10,29,45,25,10,3,0,11,27,0,4,67,10,5700000
4,2011-09-01,89,50,9,NA,NA,NA,NA,NA,NA,Investment,Mitino,12583535.69,178473,0.194702869,0.069753361,13087,6839,9,13670,17063,10,0,NA,1,0,17,6,no,0,11,4,no,no,no,no,no,no,no,no,21155,9828,11327,28563,14680,13883,120381,60040,60341,29529,9083,20446,13087,6645,6442,13670,7126,6544,32063,16513,15550,3292,1450,1842,24934,12782,12152,458,9,51,12,124,50,201,0,9,2,459,13,24,40,130,252,4,1.489049154,0.984536582,11.56562408,0.963802007,0.179440956,0.236455018,0.078090293,0.106124506,0.451173311,15.62371,2.01708,14.31764,4.2914325,51.49719001,4,3.816044582,5.271136062,4,0.131596959,1.579163513,1.200336487,no,2.677824281,14.60650078,17.45719794,18.30943312,19.48700542,2.677824284,1,no,2.780448941,17,1.999265421,no,17.47838035,83,6.734618018,1,27.6678632,9.522537611,6.348716334,1.767612439,11.17833328,0.583024969,27.89271688,0.811275289,0.62348431,1.950316967,6.483171621,7.385520691,4.923843177,3.549557568,8.789894266,2.163735157,10.9031613,0.622271644,0.599913582,0.934273498,0.8926743,0.236455018,1.03177679,1.561504846,15.30044908,16.99067736,16.04152067,5.02969633,0.465820158,good,17.36,0.57,0,0,0,0,2,1e3,1500,1250,0,0,0,2,0,0,0,0,0,0,0,0,0,19.25,10.35,1,11000,6,80780,12,658.33,1083.33,870.83,0,3,4,5,0,0,0,0,0,0,0,3,1,28.38,6.57,2,11000,7,89492,23,673.91,1130.43,902.17,0,5,9,8,1,0,0,1,0,0,0,9,2,32.29,5.73,2,11000,7,89492,25,660,1120,890,0,5,11,8,1,0,0,1,1,0,0,13,2,20.79,3.57,4,167000,12,205756,32,718.75,1218.75,968.75,0,5,14,10,3,0,0,1,2,0,0,18,3,14.18,3.89,8,244166,22,942180,61,931.58,1552.63,1242.11,4,7,21,15,11,2,1,4,4,0,0,26,3,13100000
5,2011-09-05,77,77,4,NA,NA,NA,NA,NA,NA,Investment,Basmannoe,8398460.622,108171,0.015233744,0.037316452,5706,3240,7,6748,7770,9,0,562,4,2,25,2,no,0,10,93,no,no,no,yes,yes,no,no,no,28179,13522,14657,13368,7159,6209,68043,34236,33807,26760,8563,18197,5706,2982,2724,6748,3664,3084,15237,8113,7124,5164,2583,2581,11631,6223,5408,746,48,0,0,643,16,35,0,3,1,746,371,114,146,62,53,5,1.257186453,0.876620232,8.266305238,0.68885877,0.247901208,0.376838057,0.258288769,0.236214054,0.392870988,10.68354,2.936581,11.90391,0.853960072,10.24752087,5,1.59589817,2.156283865,113,0.071480323,0.857763874,0.820294318,no,11.61665314,1.721833675,0.046809568,0.787593311,2.578670647,1.721833683,4,no,3.133530966,10,0.084112545,yes,1.59589817,113,1.423427954,4,6.515857089,8.671015673,1.638318096,3.632640421,4.587916559,2.60941961,9.15505713,1.969737724,0.220287667,2.544696,3.975401349,3.610753828,0.307915375,1.864637406,3.779781109,1.121702845,0.991682626,0.892667526,0.429052137,0.077900959,0.810801456,0.376838057,0.378755838,0.121680643,2.584369607,1.11248589,1.800124877,1.339652258,0.026102416,excellent,3.56,4.44,15,293699,1,45000,48,702.22,1166.67,934.44,3,17,10,11,7,0,0,1,4,0,2,3,0,3.34,8.29,46,420952,3,158200,153,763.45,1272.41,1017.93,8,39,45,39,19,2,1,7,12,0,6,7,0,4.12,4.83,93,1195735,9,445900,272,766.8,1272.73,1019.76,19,70,74,72,30,6,1,18,30,0,10,14,2,4.53,5.02,149,1625130,17,564843,483,765.93,1269.23,1017.58,28,130,129,131,50,14,1,35,61,0,17,21,3,5.06,8.62,305,3420907,60,2296870,1068,853.03,1410.45,1131.74,63,266,267,262,149,57,4,70,121,1,40,77,5,8.38,10.92,689,8404624,114,3503058,2283,853.88,1411.45,1132.66,143,566,578,552,319,108,17,135,236,2,91,195,14,16331452
6,2011-09-06,67,46,14,NA,NA,NA,NA,NA,NA,Investment,Nizhegorodskoe,7506452.02,43795,0.007670134,0.486245621,2418,852,2,2514,2012,3,0,NA,0,0,7,0,no,0,6,19,yes,no,no,yes,no,no,no,no,19940,9400,10540,5291,2744,2547,29660,15793,13867,8844,2608,6236,2418,1224,1194,2514,1328,1186,5866,3035,2831,4851,2329,2522,4632,2399,2233,188,24,0,0,147,2,15,0,0,0,188,0,5,152,25,6,6,2.735883907,1.593246481,18.37816963,1.531514136,0.145954816,0.113466218,1.073495427,1.497902638,0.256487453,7.18674,0.78033,14.07514,0.375311695,4.503740339,6,0.375311695,1.407418835,6,0.189227153,2.270725835,0.612447325,no,8.296086727,0.284868107,3.519388985,4.395057477,5.645795859,0.284868136,4,no,1.478528507,3,0.244670412,no,5.070196504,5,6.682088764,4,3.95950924,8.757686082,0.193126987,2.34156168,1.272894442,1.438003448,5.374563767,3.447863628,0.81041306,1.911842782,2.108923435,4.233094726,1.450974874,3.391116928,4.356122442,1.698723584,3.830021305,1.042261834,0.440707312,0.422357874,3.066285203,0.113466218,0.686931702,0.870446514,4.787705729,3.388809733,3.71355663,2.553423533,0.004469307,poor,0,19.42,5,227705,3,102000,7,1e3,1625,1312.5,3,0,1,2,1,0,0,0,0,0,0,0,0,0,40.27,10,275135,5,164000,9,883.33,1416.67,1150,3,1,1,3,1,0,0,3,1,0,0,1,0,0,50.64,18,431090,6,186400,14,718.18,1181.82,950,3,3,3,4,1,0,0,4,2,0,0,11,0,0.38,51.58,21,471290,14,683945,33,741.38,1258.62,1e3,4,5,13,8,2,1,0,6,5,0,0,21,1,1.82,39.99,54,1181009,29,1059171,120,737.96,1231.48,984.72,12,24,37,35,11,1,0,12,12,0,2,31,7,5.92,25.79,253,4274339,63,2010320,567,769.92,1280.08,1025,35,137,163,155,62,14,1,53,78,1,20,113,17,9100000
  • CSV file
  • 292 columns
  • 30k rows

Samples & Features

1 row = 1 sample = 292 columns = 291 features

price_doc: sale price (this is the target variable)
id: transaction id
timestamp: date of transaction
full_sq: total area in square meters, including loggias, balconies and other non-residential areas
life_sq: living area in square meters, excluding loggias, balconies and other non-residential areas
floor: for apartments, floor of the building
max_floor: number of floors in the building
material: wall material
build_year: year built
num_room: number of living rooms
kitch_sq: kitchen area
state: apartment condition
product_type: owner-occupier purchase or investment
sub_area: name of the district
  • numeric (276)
  • non-numeric (16)

https://php-ml.org/

PHP-ML

Machine Learning library for PHP

Load dataset

use Phpml\Dataset\CsvDataset;

$dataset = new CsvDataset('housing.csv', 291);

$dataset->getSamples();
$dataset->getTargets();

$dataset->getColumnNames();

One sample

array(291) {
  [0]=>
  string(3) "635"
  [1]=>
  string(10) "2011-12-15"
  [2]=>
  string(2) "70"
  [3]=>
  string(2) "49"
  [4]=>
  string(2) "16"
...
  string(2) "NA"
  [11]=>
  string(10) "Investment"
  [12]=>
  string(10) "Matushkino"
  [13]=>
  string(10) "4708040.47"
...

Feature selection

use Phpml\Preprocessing\ColumnFilter;

$datasetColumns = ['age', 'income', 'kids', 'beersPerWeek'];
$filterColumns = ['income', 'beersPerWeek'];

$samples = [
    [21, 100000, 1, 4],
    [35, 120000, 0, 12],
    [33, 200000, 4, 0],
];

$filter = new ColumnFilter($datasetColumns, $filterColumns);
$filter->transform($samples);

/*
[[100000, 4], [120000, 12], [200000, 0]]
*/

Number conversion

use Phpml\Preprocessing\NumberConverter;

$samples = [
    ['1', '-4'],
    ['2.0', 3.0],
    ['3', '112.5'],
    ['5', '0.0004']
];
$targets = ['1', '1', '2', '2'];

$converter = new NumberConverter();
$converter->transform($samples, $targets);
final class NumberConverter implements Preprocessor
{

    /**
     * @param mixed $nonNumericPlaceholder
     */
    public function __construct(
        bool $transformTargets = false,
        $nonNumericPlaceholder = null
    )
}

Missing values

Missing values

use Phpml\Preprocessing\Imputer;
use Phpml\Preprocessing\Imputer\Strategy\MeanStrategy;

$samples = [
    [1, null, 3, 4],
    [4, 3, 2, 1],
    [null, 6, 7, 8],
    [8, 7, null, 5],
];

$imputer = new Imputer(null, new MeanStrategy());
$imputer->fit($samples);
$imputer->transform($samples);

/* [
    [1, 2.66, 3, 4],
    [4, 3, 2, 1],
    [7, 6, 7, 8],
    [8, 7, 6.66, 5],
]*/

Model selection

use Phpml\Regression\DecisionTreeRegressor;

$model = new DecisionTreeRegressor($maxDepth = 5);
$model->train($samples, $targets);
$predictions = $model->predict($samples);

Validation

array(10) {
  [0]=>
  string(7) "5850000"
  [1]=>
  string(7) "6000000"
  [2]=>
  string(7) "5700000"
  [3]=>
  string(8) "13100000"
  [4]=>
  string(8) "16331452"
  [5]=>
  string(7) "9100000"
  [6]=>
  string(7) "5500000"
  [7]=>
  string(7) "2000000"
  [8]=>
  string(7) "5300000"
  [9]=>
  string(7) "2000000"
}
array(10) {
  [0]=>
  float(5478002.4543098)
  [1]=>
  float(4613047.2807292)
  [2]=>
  float(5478002.4543098)
  [3]=>
  float(8770788.9539153)
  [4]=>
  float(8770788.9539153)
  [5]=>
  float(6645729.5835831)
  [6]=>
  float(4613047.2807292)
  [7]=>
  float(5478002.4543098)
  [8]=>
  float(4613047.2807292)
  [9]=>
  float(4613047.2807292)
}

Train dataset != Test dataset 

Metrics

accuracy_score	 
balanced_accuracy_score	 
average_precision_score	 
brier_score_loss	 
f1_score
log_loss
precision_score
recall_score
jaccard_score
roc_auc_score

Classification

adjusted_mutual_info_score	 
adjusted_rand_score	 
completeness_score	 
fowlkes_mallows_score	 
homogeneity_score	 
mutual_info_score	 
normalized_mutual_info_score	 
v_measure_score

Clustering

explained_variance_score	 
max_error	 
mean_absolute_error	 
mean_squared_error	 
mean_squared_log_error	 
median_absolute_error	 
r2_score

Regression

This metric is best to use when targets having exponential growth, such as population counts, average sales of a commodity over a span of years etc.

MSLE: Mean squared logarithmic error

Metrics

MSLE: Mean squared logarithmic error

Best from kaggle:
 GradientBoosting

Test dataset: ~0.30087

Our model:

Train dataset: ~0.41242

Test dataset: ~0.31355

Persistence

use Phpml\ModelManager;

$modelManager = new ModelManager();
$modelManager
    ->saveToFile($model, 'housing-model.phpml');
$model = $modelManager
    ->restoreFromFile('housing-model.phpml');

$model->predict([$sample]);

Workflow

use Phpml\Regression\DecisionTreeRegressor;
use Phpml\Preprocessing\NumberConverter;
use Phpml\Pipeline;
use Phpml\Preprocessing\ColumnFilter;
use Phpml\Preprocessing\Imputer;
use Phpml\Imputer\Strategy\MeanStrategy;

$model = new Pipeline([
    new ColumnFilter(
        $trainDataset->getColumnNames(), 
        ['life_sq', 'full_sq', 'floor', 'build_year']
    ),
    new NumberConverter(true),
    new Imputer(null, new MeanStrategy()),
], new DecisionTreeRegressor(5));
$model->train($samples, $targets);

Feature union

use Phpml\FeatureUnion;

$union = new FeatureUnion([
    new Pipeline([
        new ColumnFilter($columns, ['sex']),
        new LambdaTransformer(function (array $sample) {
            return $sample[0];
        }),
        new OneHotEncoder(),
    ]),
    new Pipeline([
        new ColumnFilter($columns, ['age', 'income']),
        new NumberConverter(),
        new Imputer(null, new MeanStrategy()),
    ]),
]);

$union->fitAndTransform($samples, $targets);

Deployment

AWS Lambda

mnapoli/bref

curl -X POST \
  https://3qv1qfxlmj.execute-api.us-east-1.amazonaws.com/Prod/housing \
  -H 'content-type: multipart/form-data;' \
  -F life_sq=22 \
  -F full_sq=30 \
  -F floor=1 \
  -F build_year=2000

php-ai/php-ml-lambda-examples

Deployment

php-ai/php-ml-lambda-examples

$model = (new ModelManager())
->restoreFromFile('housing.phpml');

echo json_encode(['price' => round($model->predict([[
    $_POST['life_sq'],
    $_POST['full_sq'],
    $_POST['floor'],
    $_POST['build_year'],
]])[0])]);
$trainDataset = new CsvDataset(__DIR__.'/../data/housing-train.csv', 291);
$samples = $trainDataset->getSamples();
$targets = $trainDataset->getTargets();

$model = new Pipeline([
    new ColumnFilter($trainDataset->getColumnNames(), ['life_sq', 'full_sq', 'floor', 'build_year']),
    new NumberConverter(true),
    new Imputer(null, new MeanStrategy()),
], new DecisionTreeRegressor());

$model->train($samples, $targets);

echo 'MSLE: ' . Regression::meanSquaredLogarithmicError($targets, $model->predict($samples)) . PHP_EOL;

$testDataset = new CsvDataset(__DIR__.'/../data/housing-test.csv', 290);
$predicted = $model->predict($testDataset->getSamples());

$lines = ['id,price_doc'];
foreach ($testDataset->getSamples() as $index => $sample) {
    $lines[] = sprintf('%s,%s', $sample[0], $predicted[$index]);
}

file_put_contents(__DIR__.'/../data/housing-sub.csv', implode(PHP_EOL, $lines));

$modelManager = new ModelManager();
$modelManager->saveToFile($model, __DIR__.'/../data/housing-model.phpml');

Types of problems

Classification

5.1,3.8,1.6,0.2,setosa
4.6,3.2,1.4,0.2,setosa
5.3,3.7,1.5,0.2,setosa
5,3.3,1.4,0.2,setosa
7,3.2,4.7,1.4,versicolor
6.4,3.2,4.5,1.5,versicolor
6.9,3.1,4.9,1.5,versicolor
5.5,2.3,4,1.3,versicolor
5.9,3,5.1,1.8,virginica
5.1,3.5,1.4,0.2,setosa
4.9,3,1.4,0.2,setosa
4.7,3.2,1.3,0.2,setosa
4.6,3.1,1.5,0.2,setosa
5,3.6,1.4,0.2,setosa
5.4,3.9,1.7,0.4,setosa
4.6,3.4,1.4,0.3,setosa
5,3.4,1.5,0.2,setosa
4.4,2.9,1.4,0.2,setosa
4.9,3.1,1.5,0.1,setosa
5.4,3.7,1.5,0.2,setosa
4.8,3.4,1.6,0.2,setosa
4.8,3,1.4,0.1,setosa
4.3,3,1.1,0.1,setosa
5.8,4,1.2,0.2,setosa
5.7,4.4,1.5,0.4,setosa
5.4,3.9,1.3,0.4,setosa
5.1,3.5,1.4,0.3,setosa
5.7,3.8,1.7,0.3,setosa
5.1,3.8,1.5,0.3,setosa
5.4,3.4,1.7,0.2,setosa
5.1,3.7,1.5,0.4,setosa
4.6,3.6,1,0.2,setosa
5.1,3.3,1.7,0.5,setosa
4.8,3.4,1.9,0.2,setosa
5,3,1.6,0.2,setosa
5,3.4,1.6,0.4,setosa
5.2,3.5,1.5,0.2,setosa
5.2,3.4,1.4,0.2,setosa
4.7,3.2,1.6,0.2,setosa
4.8,3.1,1.6,0.2,setosa
5.4,3.4,1.5,0.4,setosa
5.2,4.1,1.5,0.1,setosa
5.5,4.2,1.4,0.2,setosa
4.9,3.1,1.5,0.1,setosa
5,3.2,1.2,0.2,setosa
5.5,3.5,1.3,0.2,setosa
4.9,3.1,1.5,0.1,setosa
4.4,3,1.3,0.2,setosa
5.1,3.4,1.5,0.2,setosa
5,3.5,1.3,0.3,setosa
4.5,2.3,1.3,0.3,setosa
4.4,3.2,1.3,0.2,setosa
5,3.5,1.6,0.6,setosa
5.1,3.8,1.9,0.4,setosa
4.8,3,1.4,0.3,setosa
5.1,3.8,1.6,0.2,setosa
4.6,3.2,1.4,0.2,setosa
5.3,3.7,1.5,0.2,setosa
5,3.3,1.4,0.2,setosa
7,3.2,4.7,1.4,versicolor
6.4,3.2,4.5,1.5,versicolor
6.9,3.1,4.9,1.5,versicolor
5.5,2.3,4,1.3,versicolor
6.5,2.8,4.6,1.5,versicolor
5.7,2.8,4.5,1.3,versicolor
6.3,3.3,4.7,1.6,versicolor
4.9,2.4,3.3,1,versicolor
6.6,2.9,4.6,1.3,versicolor
5.2,2.7,3.9,1.4,versicolor
5,2,3.5,1,versicolor
5.9,3,4.2,1.5,versicolor
6,2.2,4,1,versicolor
6.1,2.9,4.7,1.4,versicolor
5.6,2.9,3.6,1.3,versicolor
6.7,3.1,4.4,1.4,versicolor
5.6,3,4.5,1.5,versicolor
5.8,2.7,4.1,1,versicolor
6.2,2.2,4.5,1.5,versicolor
5.6,2.5,3.9,1.1,versicolor
5.9,3.2,4.8,1.8,versicolor
6.1,2.8,4,1.3,versicolor
6.3,2.5,4.9,1.5,versicolor
6.1,2.8,4.7,1.2,versicolor
6.4,2.9,4.3,1.3,versicolor
6.6,3,4.4,1.4,versicolor
6.8,2.8,4.8,1.4,versicolor
6.7,3,5,1.7,versicolor
6,2.9,4.5,1.5,versicolor
5.7,2.6,3.5,1,versicolor
5.5,2.4,3.8,1.1,versicolor
5.5,2.4,3.7,1,versicolor
5.8,2.7,3.9,1.2,versicolor
6,2.7,5.1,1.6,versicolor
5.4,3,4.5,1.5,versicolor
6,3.4,4.5,1.6,versicolor
6.7,3.1,4.7,1.5,versicolor
6.3,2.3,4.4,1.3,versicolor
5.6,3,4.1,1.3,versicolor
5.5,2.5,4,1.3,versicolor
5.5,2.6,4.4,1.2,versicolor
6.1,3,4.6,1.4,versicolor
5.8,2.6,4,1.2,versicolor
5,2.3,3.3,1,versicolor
5.6,2.7,4.2,1.3,versicolor
5.7,3,4.2,1.2,versicolor
5.7,2.9,4.2,1.3,versicolor
6.2,2.9,4.3,1.3,versicolor
5.1,2.5,3,1.1,versicolor
5.7,2.8,4.1,1.3,versicolor
6.3,3.3,6,2.5,virginica
5.8,2.7,5.1,1.9,virginica
7.1,3,5.9,2.1,virginica
6.3,2.9,5.6,1.8,virginica
6.5,3,5.8,2.2,virginica
7.6,3,6.6,2.1,virginica
4.9,2.5,4.5,1.7,virginica
7.3,2.9,6.3,1.8,virginica
6.7,2.5,5.8,1.8,virginica
7.2,3.6,6.1,2.5,virginica
6.5,3.2,5.1,2,virginica
6.4,2.7,5.3,1.9,virginica
6.8,3,5.5,2.1,virginica
5.7,2.5,5,2,virginica
5.8,2.8,5.1,2.4,virginica
6.4,3.2,5.3,2.3,virginica
6.5,3,5.5,1.8,virginica
7.7,3.8,6.7,2.2,virginica
7.7,2.6,6.9,2.3,virginica
6,2.2,5,1.5,virginica
6.9,3.2,5.7,2.3,virginica
5.6,2.8,4.9,2,virginica
7.7,2.8,6.7,2,virginica
6.3,2.7,4.9,1.8,virginica
6.7,3.3,5.7,2.1,virginica
7.2,3.2,6,1.8,virginica
6.2,2.8,4.8,1.8,virginica
6.1,3,4.9,1.8,virginica
6.4,2.8,5.6,2.1,virginica
7.2,3,5.8,1.6,virginica
7.4,2.8,6.1,1.9,virginica
7.9,3.8,6.4,2,virginica
6.4,2.8,5.6,2.2,virginica
6.3,2.8,5.1,1.5,virginica
6.1,2.6,5.6,1.4,virginica
7.7,3,6.1,2.3,virginica
6.3,3.4,5.6,2.4,virginica
6.4,3.1,5.5,1.8,virginica
6,3,4.8,1.8,virginica
6.9,3.1,5.4,2.1,virginica
6.7,3.1,5.6,2.4,virginica
6.9,3.1,5.1,2.3,virginica
5.8,2.7,5.1,1.9,virginica
6.8,3.2,5.9,2.3,virginica
6.7,3.3,5.7,2.5,virginica
6.7,3,5.2,2.3,virginica
6.3,2.5,5,1.9,virginica
6.5,3,5.2,2,virginica
6.2,3.4,5.4,2.3,virginica
5.9,3,5.1,1.8,virginica

Classification

use Phpml\Classification\KNearestNeighbors;
use Phpml\CrossValidation\RandomSplit;
use Phpml\Dataset\Demo\IrisDataset;
use Phpml\Metric\Accuracy;

$dataset = new IrisDataset();
$split = new RandomSplit($dataset);

$classifier = new KNearestNeighbors();
$classifier->train(
    $split->getTrainSamples(), 
    $split->getTrainLabels()
);

$predicted = $classifier->predict($split->getTestSamples());

echo sprintf("Accuracy: %s", 
    Accuracy::score($split->getTestLabels(), $predicted)
);

Classification

51.403007;7.208546;good
52.2688736;10.5267696;good
52.235083;5.181552;good
47.5292;9.9267;good
47.9704873;17.7852936;moderate
51.4625;13.526666666667;moderate
48.7344444;19.1128073;good
46.815277777778;9.8558333333333;good
51.1506269;14.968707;good
47.041668;15.433056;good
50.303207830366;6.0017362891249;good
47.178333333333;14.676666666667;good
47.102253726073;9.3375502158063;good
47.5818083;12.1724111;good
50.97;9.8;moderate
47.146155;5.551039;good
51.64265556;15.12780833;good
54.353333333333;18.635277777778;good
50.5425641;12.7792228;good
52.234504;6.919494;moderate
50.467528;13.412696;good
51.233652208391;5.1639788468472;good
52.14325;19.233225;good
45.14254358;10.04384767;unhealthy for sensitive
49.228472;17.675083;unhealthy for sensitive
50.80425833;8.76932778;moderate
48.396866667006;9.9789750003815;moderate
53.4708393;7.4848308;good
47.871158;17.273464;good
48.33472;16.729445;good
47.409443;15.253333;good

Visualization

$minLat = 41.34343606848294;
$maxLat = 57.844750992891;
$minLng = -16.040039062500004;
$maxLng = 29.311523437500004;

$step = 0.1;
$k = 3;

$dataset = new CsvDataset(__DIR__.'/../data/air.csv', 2, false, ';');
$estimator = new KNearestNeighbors($k);
$estimator->train($dataset->getSamples(), $dataset->getTargets());

$lines = [];
for ($lat=$minLat; $lat<$maxLat; $lat+=$step) {
    for ($lng=$minLng; $lng<$maxLng; $lng+=$step) {
        $lines[] = sprintf('%s;%s;%s', $lat, $lng, $estimator->predict([[$lat, $lng]])[0]);
    }
}

file_put_contents(__DIR__.'/../data/airVis.csv', implode(PHP_EOL, $lines));

Regression

Miles	Price
9300	7100
10565	15500
15000	4400
15000	4400
17764	5900
57000	4600
65940	8800
73676	2000
77006	2750
93739	2550
146088	960
153260	1025

Regression

use Phpml\Regression\LeastSquares;

$samples = [[9300], [10565], [15000], [15000], [17764], [57000], [65940], [73676], [77006], [93739], [146088], [153260]];
$targets = [7100, 15500, 4400, 4400, 5900, 4600, 8800, 2000, 2750, 2550,  960, 1025];

$regression = new LeastSquares();
$regression->train($samples, $targets);


$regression->getCoefficients();
$regression->getIntercept();

$price = $regression->predict([35000]);

Clustering

41.793935909;-87.625680278
41.907274031;-87.722791892
41.851296671;-87.706458801
41.775963639;-87.615517372
41.794879;-87.63179049
41.799461412;-87.596206318
41.989599401;-87.660256868
42.019398729;-87.67543958
42.004487311;-87.679846425
42.009087258;-87.690171862
41.799518433;-87.590997844
41.875039579;-87.743690267
41.875198392;-87.717479393
41.78640901;-87.649813179
41.766229647;-87.577855722
41.900062396;-87.620884259
41.744708666;-87.616371298
41.7737319;-87.651916442
41.692289426;-87.647852131
41.874236291;-87.674657583
41.761450225;-87.623211368
41.831030756;-87.624424247
41.974853031;-87.713545123
41.974605662;-87.660819291
41.815419529;-87.702711186
41.750341521;-87.657371388
41.854659562;-87.716303651
41.834650408;-87.62843175
41.793435216;-87.70876482
41.894904052;-87.626344479
41.894993069;-87.746918939
41.90984267;-87.729545576
41.967477901;-87.739224006
41.87522978;-87.728549617
41.765946803;-87.595563723
41.908222431;-87.679234761
41.882757453;-87.709603286
41.876121224;-87.641003973
41.809372853;-87.704024967
41.977043475;-87.76899404
41.943664148;-87.646353396
41.759350571;-87.623168543
41.693840666;-87.613406835
41.964351639;-87.661000678

Clustering

Clustering

Clustering

Clustering

use Phpml\Clustering\KMeans;

$samples = [
    [1, 1], [8, 7], [1, 2], 
    [7, 8], [2, 1], [8, 9]
];

$kmeans = new KMeans(2);
$clusters = $kmeans->cluster($samples);

$clusters = [
    [[1, 1], [1, 2], [2, 1]],
    [[8, 7], [7, 8], [8, 9]]
];

Feature Extraction

use Phpml\FeatureExtraction\TokenCountVectorizer;
use Phpml\Tokenization\WhitespaceTokenizer;

$samples = [
    'Lorem ipsum dolor sit amet dolor',
    'Mauris placerat ipsum dolor',
    'Mauris diam eros fringilla diam',
];

$vectorizer = new TokenCountVectorizer(new WhitespaceTokenizer());

$vectorizer->fit($samples);
$vectorizer->getVocabulary()


$vectorizer->transform($samples);

$tokensCounts = [
    [0 => 1, 1 => 1, 2 => 2, 3 => 1, 4 => 1, 5 => 0, 6 => 0, 7 => 0, 8 => 0, 9 => 0],
    [0 => 0, 1 => 1, 2 => 1, 3 => 0, 4 => 0, 5 => 1, 6 => 1, 7 => 0, 8 => 0, 9 => 0],
    [0 => 0, 1 => 0, 2 => 0, 3 => 0, 4 => 0, 5 => 1, 6 => 0, 7 => 2, 8 => 1, 9 => 1],
];

Model selection

use Phpml\CrossValidation\RandomSplit;
use Phpml\CrossValidation\StratifiedRandomSplit;
use Phpml\Dataset\ArrayDataset;


$dataset = new ArrayDataset(
    $samples = [[1], [2], [3], [4]],
    $labels = ['a', 'a', 'b', 'b']
);

$randomSplit = new RandomSplit($dataset, 0.5);
$dataset = new ArrayDataset(
    $samples = [[1], [2], [3], [4], [5], [6], [7], [8]],
    $labels = ['a', 'a', 'a', 'a', 'b', 'b', 'b', 'b']
);

$split = new StratifiedRandomSplit($dataset, 0.5);
  • Feature Selection
  • Dimensionality Reduction
  • Datasets 
  • Models Management
  • Neural Network
  • Metric
  • Association Rule Learning
  • Ensemble Algorithms
  • Math

https://github.com/php-ai/php-ml

Summary

ML is all about the proccess

  • Define a problem
  • Gather your data
  • Prepare your data for ML
  • Select algorithm
  • Train model
  • Tune parameters
  • Select final model
  • Validate final model
  • Deploy and maintenance

source: http://scikit-learn.org/stable/tutorial/machine_learning_map/index.html

Summary

  • the most important is question, data is secondary (but also important)
  • many algorithms and techniques
  • application can be very simple but also extremely sophiticated
  • sometimes difficult to find correct answer
  • base math skills are important

Summary

https://twitter.com/JustinMatejka/status/859075295059562498

PHP 7.0
206,128 instances classified in 30 seconds 
(6,871 per second)

https://github.com/syntheticminds/php7-ml-comparison

Python 2.7
106,879 instances classified in 30 seconds 
(3,562 per second)
NodeJS v5.11.1
245,227 instances classified in 30 seconds 
(8,174 per second)
Java 8
16,809,048 instances classified in 30 seconds 
(560,301 per second)
PHP 7.1
302,931 instances classified in 30 seconds
(10,098 per second)
PHP 7.2
365,568 instances classified in 30 seconds
(12,186 per second)
PHP 7.3
408,667 instances classified in 30 seconds
(13,622 per second)

Event Storming

Where to begin?

kaggle.com

What to read?

Q&A

Thanks for listening

@ ArkadiuszKondas

https://arkadiuszkondas.com

https://php-ml.org/

https://slides.com/arkadiuszkondas/machine-learning-in-php-2019/

https://joind.in/talk/98e38

Machine Learning - how to start to teach the machine - PHP Russia 2019

By Arkadiusz Kondas

Machine Learning - how to start to teach the machine - PHP Russia 2019

The main goal of Machine Learning is to create intelligent systems that can improve and acquire new knowledge using input data. In practice, this translates into the use of one of hundreds of different available algorithms. This lecture is an introduction to ML from the total basics. We will learn the basic vocabulary and types of problems which the ML allows. I will also present the technique of building the whole pipeline, with the help of which we will go through all stages of ML: data processing (preprocessing), selection of algorithms and evaluation of its effectiveness.

  • 251
Loading comments...

More from Arkadiusz Kondas