我有一个名为“cleaned”的数据框，它由大约 300,000 行和 13 个变量组成。除因变量外，所有变量都是分类变量，具有多个水平（）。因变量是数字，取值范围为 -1,500 到 3,296，大部分为正。以下是数据框的摘要：
> summary(cleaned$SumOf1st.Yr.Cash)
Min. 1st Qu. Median Mean 3rd Qu. Max.
-1574.00 37.37 101.50 155.60 204.60 3296.00
> dput(head(cleaned))
structure(list(Submit.Qtr = structure(c(2L, 2L, 2L, 2L, 2L, 2L
), .Label = c("1Q11", "2Q11", "3Q11", "4Q11"), class = "factor"),
SUBMIT_CAL_MONTH = c(201104L, 201104L, 201104L, 201104L,
201104L, 201104L), SumOf1st.Yr.Cash = c(-221.81, 127.86,
662.09, 77.24, 370.4, 176), CARRIER_NAME = structure(c(56L,
116L, 4L, 116L, 82L, 114L), .Label = c("AARP-branded plans, insured by Aetna",
"Aetna", "Aetna Life Insurance Company", "Altius Health Plans",
"Altius One", "American Family Life Assurance Company of Columbus (Aflac)",
"American Family Life Assurance Company of New York (Aflac New York)",
"AmeriHealth - New Jersey", "Ameritas Life Insurance Corp.",
"Anthem BCBS. Serving residents of Indiana", "Anthem BCBS. Serving residents of Kentucky",
"Anthem BCBS. Serving residents of Ohio", "Anthem Blue Cross",
"Anthem Blue Cross and Blue Shield", "Anthem Blue Cross and Blue Shield of CT",
"Anthem Blue Cross and Blue Shield of NH", "Anthem Blue Cross and Blue Shield of VA",
"Anthem Blue Cross Blue Shield", "Anthem Blue Cross Blue Shield Indiana",
"Anthem Blue Cross Blue Shield Kentucky", "Anthem Blue Cross Blue Shield of Connecticut",
"Anthem Blue Cross Blue Shield of Missouri", "Anthem Blue Cross Blue Shield of Wisconsin",
"Anthem Blue Cross Blue Shield Ohio", "Anthem BlueCross BlueShield",
"Anthem Health Plans of Kentucky Inc.", "Anthem Health Plans of New Hampshire Inc.",
"Argus", "Arise Health Plan", "Arkansas Blue Cross and Blue Shield",
"Assurant", "Assurant Employee Benefits", "Assurant Health",
"Asuris Northwest Health", "Avera Health Plans", "AvMed Health Plans",
"Bay Dental", "BCBS of GA", "Blue Cross and Blue Shield of GA",
"Blue Cross and Blue Shield of Georgia", "Blue Cross and Blue Shield of Illinois",
"Blue Cross and Blue Shield of Kansas City", "Blue Cross and Blue Shield of Minnesota",
"Blue Cross and Blue Shield of South Carolina", "Blue Cross and Blue Shield of Texas",
"Blue Cross Blue Shield", "Blue Cross Blue Shield of Arizona",
"Blue Cross Blue Shield of Delaware", "Blue Cross Blue Shield of Florida",
"Blue Cross Blue Shield of Georgia", "Blue Cross Blue Shield of Michigan",
"Blue Cross Blue Shield of North Dakota", "Blue Cross of Idaho",
"Blue Cross of Northeastern Pennsylvania through its subsidiary First Priority Life Insurance Company",
"Blue Shield of California", "BlueCross BlueShield of Louisiana",
"BlueCross BlueShield of Montana", "BlueCross BlueShield of Nebraska",
"BlueCross BlueShield of Tennessee", "BlueCross BlueShield of Wyoming",
"Capital Blue Cross", "Care Improvement Plus", "CareFirst BlueCross BlueShield",
"Celtic Ins. Co.", "CeltiCare Healthplan of MA, Inc.", "Cigna",
"Clear One Health Plans", "ConnectiCare Inc.", "Coventry",
"Coventry Health and Life Insurance Co. FL", "Coventry Health and Life Insurance Company",
"Coventry Health Care of Delaware, Inc", "Coventry Health Care of Georgia, Inc.",
"Coventry Health Care of Illinois, Inc.", "Coventry Health Care of Iowa, Inc.",
"Coventry Health Care of Kansas Inc.", "Coventry Health Care of Louisiana, Inc.",
"Coventry Health Care of Missouri, Inc.", "Coventry Health Care of Oklahoma Inc.",
"Coventry Health Care of the Carolinas, Inc.", "Coventry Health Plan of Florida, Inc.",
"Dean Health Plan, Inc.", "Delta Dental Insurance Company (Delta Dental)",
"Delta Dental of California", "Delta Dental of Colorado",
"Delta Dental of Iowa", "Delta Dental of Minnesota", "Delta Dental of North Carolina",
"Dentegra Insurance Company", "Dominion Dental Services, Inc",
"Easy Choice Health Plan of New York", "EmblemHealth", "Empire",
"Empire BlueCross", "Evercare by UnitedHealthcare", "Everest Dental Plan",
"Fallon Community Health Plan", "Geisinger Choice", "Generic Medicare Carrier",
"Group Health", "HCC Life Insurance Company", "HCC Medical Insurance Services",
"Health Alliance Plan", "Health Insurance Innovations", "Health Net",
"Health Net of Arizona", "Health Net of Oregon", "Health Plan of Nevada",
"HealthAmerica", "HealthPartners", "HealthPlus Insurance Company",
"Highmark Blue Cross Blue Shield Delaware", "Highmark Blue Cross Blue Shield West Virginia",
"Horizon Blue Cross Blue Shield of New Jersey", "Humana",
"Humana CompBenefits", "Humana Health Benefit Plan of Louisiana Inc.",
"Humana Health Insurance Company of Florida", "Humana Insurance Company of Kentucky",
"IHC Group", "IMG Global", "Independence Blue Cross", "Kaiser Foundation Health Plan of the NW",
"Kaiser Mid-Atlantic", "Kaiser Permanente CO", "Kaiser Permanente GA",
"Kaiser Permanente of CA", "Kaiser Permanente of HI", "Kaiser Permanente of Ohio",
"KPS Health Plans", "LifeWise Health Plan of Oregon", "LifeWise Health Plan of Washington",
"Lovelace Health Plans", "Madison National Life Insurance Company",
"Medica", "Medica of Minnesota", "Medical Mutual", "Mercy Health Plans",
"Mutual of Omaha", "Mutual Of Omaha", "Mutual of Omaha Insurance Company",
"MVP", "My Health Alliance", "Nationwide Life Insurance Company",
"Next Generation Insurance Group", "ODS Alaska", "ODS Health Plan, Inc.",
"Optima Health Insurance Company", "Oxford NJ", "Oxford NY",
"PacifiCare", "PacificSource Health Plans", "PacificSource Health Plans of Idaho",
"Physicians Health Plan of Northern Indiana, Inc.", "Physicians Plus",
"PreferredOne Insurance Company", "Premera Blue Cross", "Premera Blue Cross Blue Shield of Alaska",
"Presbyterian", "Providence Health Plan", "QCA Health Plan Inc",
"Regence Blue Cross Blue Shield of Oregon", "Regence Blue Cross Blue Shield of Utah",
"Regence Blue Shield of Idaho", "Regence BlueCross BlueShield of Oregon",
"Regence BlueCross BlueShield of Utah", "Regence BlueShield",
"Regence BlueShield of Idaho", "Regence Life and Health",
"Regence Life and Health Insurance Company", "RegenceBCBS",
"RegenceBS", "Rocky Mountain Health Plans", "Scott & White Health Plan",
"SecureHorizons by UnitedHealthcare", "Security Health Plan",
"Security Life Insurance Company of America", "SelectHealth",
"Seven Corners", "Sierra Health and Life", "Standard Security Life",
"Standard Security Life Insurance Company", "SummaCare Inc of Ohio",
"Symetra Life Insurance Company", "Total Dental Administrators Health Plan, Inc.",
"UniCare", "United Concordia Dental", "United of Omaha",
"United World", "UnitedHealthcare", "UnitedHealthcare Community Plan",
"UnitedHealthOne", "Unity Health Insurance", "Vision Plan of America",
"VSP", "WellCare", "WellCare Health Plans of New Jersey, Inc.",
"WellCare of Florida, Inc.", "WellCare of New York, Inc.",
"WellCare of Ohio, Inc.", "WellCare of Texas, Inc.", "WellCare Prescription Insurance, Inc.",
"Wellmark Blue Cross and Blue Shield of Iowa", "Wellmark Blue Cross and Blue Shield of South Dakota",
"WellPath Select, Inc.", "WINhealth Partners", "WPS", "WPS Health Insurance"
), class = "factor"), GENDER = structure(c(2L, 2L, 1L, 1L,
1L, 1L), .Label = c("F", "M", "U"), class = "factor"), FAMILY_TYPE = structure(c(1L,
1L, 1L, 1L, 2L, 2L), .Label = c("FAMILY", "INDIVIDUAL", "Unknown"
), class = "factor"), EXECUTIVE_AGE_GROUP = structure(c(5L,
5L, 4L, 3L, 3L, 6L), .Label = c("0-18 (>=0 AND <19)", "19-25 (>=19 AND <26)",
"26-29 (>=26 AND <30)", "30-39 (>=30 AND <40)", "40-49 (>=40 AND <50)",
"50-64 (>=50 AND <65)", "65+", "Unknown"), class = "factor"),
MARITAL_STATUS = structure(c(4L, 1L, 8L, 1L, 7L, 1L), .Label = c("",
"D", "L", "M", "O", "P", "S", "W"), class = "factor"), SELECTED_RIDERS = structure(c(11L,
1L, 1L, 10L, 1L, 10L), .Label = c("", "/Dental_Vision/",
"/Dental/", "/Dental/Other/", "/Life/", "/Life/Dental_Vision/",
"/Life/Dental/", "/Life/Dental/Other/", "/Life/Other/", "/None/",
"/Other/", "/Vision/", "/Vision/Dental/", "/Vision/Dental/Other/",
"/Vision/Life/", "/Vision/Life/Dental/", "/Vision/Life/Dental/Other/",
"/Vision/Life/Other/", "/Vision/Other/"), class = "factor"),
STATE_ABBR = structure(c(19L, 44L, 45L, 10L, 49L, 32L), .Label = c("AK",
"AL", "AR", "AZ", "CA", "CO", "CT", "DC", "DE", "FL", "GA",
"HI", "IA", "ID", "IL", "IN", "KS", "KY", "LA", "MA", "MD",
"ME", "MI", "MN", "MO", "MS", "MT", "NC", "ND", "NE", "NH",
"NJ", "NM", "NV", "NY", "OH", "OK", "OR", "PA", "RI", "SC",
"SD", "TN", "TX", "UT", "VA", "VT", "WA", "WI", "WV", "WY"
), class = "factor"), SumOfAPPROVED.MEMBERS = c(7L, 6L, 4L,
2L, 1L, 1L), PRODUCTLINE_TYPE = structure(c(4L, 3L, 4L, 3L,
4L, 4L), .Label = c("ACC", "CAN", "DT", "IFP", "MA", "MAPD",
"MD", "MS", "PDC", "ST", "STU", "TRV", "VSP"), class = "factor"),
CHANNEL = structure(c(1L, 1L, 2L, 3L, 2L, 3L), .Label = c("Direct",
"Online Advertising", "Performance Partners"), class = "factor")), .Names = c("Submit.Qtr",
"SUBMIT_CAL_MONTH", "SumOf1st.Yr.Cash", "CARRIER_NAME", "GENDER",
"FAMILY_TYPE", "EXECUTIVE_AGE_GROUP", "MARITAL_STATUS", "SELECTED_RIDERS",
"STATE_ABBR", "SumOfAPPROVED.MEMBERS", "PRODUCTLINE_TYPE", "CHANNEL"
), row.names = c(2L, 4L, 5L, 6L, 7L, 8L), class = "data.frame")
我试图对这些数据运行混合效应模型
lme(SumOf1st.Yr.Cash~ GENDER + FAMILY_TYPE +
+ EXECUTIVE_AGE_GROUP + MARITAL_STATUS + SELECTED_RIDERS +
+ PRODUCTLINE_TYPE + CHANNEL, data=cleaned, random= ~1 | STATE_ABBR)
但是我收到一条我无法理解的错误消息。
Error in MEEM(object, conLin, control$niterEM) :
Singularity in backsolve at level 0, block 1
当公式中只有一个变量时，我可以运行它。但即使有两个，我也收到了相同的错误消息。
lme(SumOf1st.Yr.Cash~ GENDER +
+ EXECUTIVE_AGE_GROUP , data=cleaned, random= ~1 | STATE_ABBR)
Error in MEEM(object, conLin, control$niterEM) :
Singularity in backsolve at level 0, block 1
lme(SumOf1st.Yr.Cash~ GENDER
+ , data=cleaned, random= ~1 | STATE_ABBR)
Linear mixed-effects model fit by REML
Data: cleaned
Log-restricted-likelihood: -1877607
Fixed: SumOf1st.Yr.Cash ~ GENDER
(Intercept) GENDERM GENDERU
146.53490 12.27621 494.77983
Random effects:
Formula: ~1 | STATE_ABBR
(Intercept) Residual
StdDev: 39.21808 177.8023
Number of Observations: 284485
Number of Groups: 51