四川移动客户细分模型

单击此处编辑母版标题样式,单击此处编辑母版文本样式,第二级,第三级,第四级,第五级,*,*,2003年12月,2003,年,年 12,月,月,四川移动客,户,户细分模型,项,项目报告,1,报告会议程,一、项目背,景,景及前阶段,工,工作回顾,二、客户细,分,分方法论,三、四川移,动,动客户细分,模,模型,四、通用客,户,户细分结果,分,分析及相应,市,市场策略,五、短信专,题,题客户细分,结,结果分析及,相,相应市场策,略,略,六、基于客,户,户细分结果,的,的市场营销,活,活动讨论,2,报告会议程,一、项目背,景,景及前阶段,工,工作回顾,二、客户细,分,分方法论,三、四川移,动,动客户细分,模,模型,四、通用客,户,户细分结果,分,分析及相应,市,市场策略,五、短信专,题,题客户细分,结,结果分析及,相,相应市场策,略,略,六、基于客,户,户细分结果,的,的市场营销,活,活动讨论,3,项目概述,项目主题：,四川移动客,户,户细分模型,项目范围：,本期项目以,乐,乐山市为试,点,点城市,项目目的：,按用户行为,进,进行细分，,客,客观反映用,户,户需求。通,过,过对各类人,群,群的深入分,析,析，为相关,部,部门制订资,费,费、服务、,市,市场策略提,供,供基础。,项目内容：,建立客户细,分,分模型,结合各部门,需,需求对客户,细,细分群进行,详,详细分析,协助市场经,营,营部进行相,关,关服务、市,场,场活动的策,划,划,4,项目各阶段,计,计划,第一阶段：,需,需求调研、,项,项目范围界,定,定,在与各部门,的,的访谈中，,了,了解各部门,的,的需求点，,以,以便在客户,细,细分后有重,点,点地进行分,析,析。,第二阶段：,收,收集提取数,据,据,依据第一阶,段,段的需求，,收,收集并整理,数,数据，把数,据,据加载到指,定,定的数据库,表,表，为下一,阶,阶段的建模,提,提供基础。,第三阶段：,建,建模及分析,建立细分模,型,型，对乐山,移,移动用户进,行,行细分。并,根,根据前期访,谈,谈所了解的,需,需求，结合,乐,乐山具体情,况,况，对客户,细,细分群进行,详,详细分析。,第四阶段：,应,应用专题设,计,计,根据第一阶,段,段的访谈，,并,并在第三阶,段,段的细分结,果,果分析基础,上,上，协助市,场,场经营部门,对,对部分重点,细,细分群体有,针,针对性地设,计,计市场策略,和,和市场活动,。,。,注：SPSS的CRISP建模流,程,程包含了第,二,二、三阶段,，,，所以此阶,段,段内容在建,模,模步骤中详,细,细讲述。,5,150300元（市州）,500元以下（成都）,普通全球通客户,300元（市州）,500元（成都）,贵宾卡,800元,全球通银卡,1500元,全球通金卡,3000元,全球通钻石卡,ARPU,全球通客户等级,四川移动已,对,对用户按ARPU值进,行,行了用户细,分,分，并对不,同,同级别的客,户,户采取了不,同,同的服务策,略,略。,四川移动客,户,户细分现状,6,高端用户,中端用户,低端用户,以全球通新,形,形象涵盖目,标,标群的共性,特,特质,以顶,级,级服,务,务/,享,享受,满,满足,和,和稳,固,固,以领,先,先业,务,务刺,激,激和,带,带动,目前,客,客户,细,细分,无,无法,完,完全,满,满足,业,业,务,务部,门,门增,长,长需,求,求,通过,业,业务,访,访谈,，,，我,们,们了,解,解到,基,基于ARPU,值,值的,客,客户,细,细分,还,还无,法,法完,全,全满,足,足业,务,务部,门,门日,益,益增,长,长的,业,业务,需,需求,准确,了,了解,客,客户,行,行为,可,可以,进,进一,步,步了,解,解客,户,户，,增,增加,客,客户,满,满意,度,度,7,针对,各,各部,门,门的,需,需求,访,访谈,访谈部门,需求点,策略分析,计费业务中心,主要通过计费业务中心了解客户细分项目系统平台的需求和系统环境,市场经营部,关注全球通品牌整合，希望通过品牌整合提高服务品质，实现在业务、回馈、渠道上的差异化,通过客户细分，分析品牌整合高中端人群行为消费特征，为其设计差异化服务。并对品牌整合的效果从客户细分角度进行监测。,如何更好服务于中端用户,在客户细分后准确定位全球通中端用户，研究其行为方式和特征，进行有针对性的服务设计和活动策划。,通过提高服务来挽留有离网倾向用户,研究分析有价值的离网用户行为特征，策划有针对性的活动或服务进行挽留。,营销中心,关注双机倾向用户的挽留。希望通过发现有双机倾向的用户，并在其购买之前进行挽留,分析以往双机用户的行为特征及关注点，针对有价值用户设计服务项目及策划活动。,预警离网用户，通过提高服务来挽留有离网倾向用户,研究分析有价值的离网用户行为特征，策划有针对性的活动或服务进行挽留。,关注大众卡的效果,通过客户细分了解大众卡的受众群，向有流失可能的用户进行宣传，并避免直接影响全球通其他用户。,通过,对,对各,部,部门,的,的访,谈,谈，,为,为客,户,户细,分,分模,型,型的,建,建立,提,提供,指,指导,依,依据,。,。,8,针对,各,各部,门,门的,需,需求,访,访谈,访谈部门,需求点,策略分析,客户服务中心客户关系管理室,目前的外呼服务多为服务宣传和用户调查。外呼用户多为随机抽取，缺乏策略性。希望通过客户细分节约外呼成本，提高外呼效率和效果，并深入挖掘外呼应用。,对客户进行细分后，在外呼时可提供有针对性的服务。在推荐产品或进行调查前便了解用户的行为特征、生活方式等，有助于提高亲和力，节省人员和资金成本，并提升效果。,数据部,对于目标用户只了解其大致特征，不清楚其更细致的行为特征、生活方式等，因此在推各项业务时定位不准。希望通过客户细分模型了解各种数据业务的使用人群以及使用行为方式。,通过客户细分，寻找其重要业务的目标人群，并详细分析特征，对业务的规则制定、推广提供依据。,乐山市场部,小灵通和联通竞争激烈，希望了解有转网倾向的用户的特征，并进行及时挽留,分析有可能转至小灵通、联通的用户，由于整个样本空间规模较小，可以适当对中高端人群制定一对一的服务策略,目前所有产品都针对所有人群进行推广，希望找到目标人群有的放矢地进行推广，提高效率，节约成本,。,分析各种产品的目标用户，研究其行为特征，并通过交叉销售带动整个产品的销售增量。,希望找到短信密集使用人群，并以此为导入点，制定有效的方式刺激整个短信市场,通过对短信密集人群的行为分析，找到此类人群，通过他们带动周围人群的短信使用量，并通过他们对移动产品及业务进行软性宣传,9,报告,会,会议,程,程,一、,项,项目,背,背景,及,及前,阶,阶段,工,工作,回,回顾,二、,客,客户,细,细分,方,方法,论,论,三、,四,四川,移,移动,客,客户,细,细分,模,模型,四、,通,通用,客,客户,细,细分,结,结果,分,分析,及,及相,应,应市,场,场策,略,略,五、,短,短信,专,专题,客,客户,细,细分,结,结果,分,分析,及,及相,应,应市,场,场策,略,略,六、,基,基于,客,客户,细,细分,结,结果,的,的市,场,场营,销,销活,动,动讨,论,论,10,客户,细,细分,分,分析,结,结果,能,能帮,助,助四,川,川,移,移动,实,实现,股,股东,价,价值,的,的最,大,大化,从各,个,个角,度,度来,增,增加,股,股东,价,价值,，,，都,需,需要,以,以客,户,户细,分,分为,基,基础,。,。,股东价值最大化,利润增长,收入增长,资产回报增长,增加现有的收入,(,现有客户),创造新的收入,(,新客户),提高客户接触成本有效性,提高资产收益率,增加客户周期,（降低客户流失率,),提高现在的客户关系,(增加现有产/服务销售),提高现在的客户关系,(增加新产品/服务销售),提高在现有的客户群中的份额,进入新的客户群,提供新产品和服务,提高营销成本有效性,提高销售成本有效性,提高服务成本有效性,提高库存/,定单管理成本有效性,提高基础设施/中央服务设施成本,有效性,将资源分配给高价值的产品/客户,提升客户细分/区分度,进行产品和渠道的重新组合,客户群信息,德勤咨询专有的客户价值图,进入新的市场,（地理上）,11,客户,细,细分,是,是产,品,品开,发,发、,市,市场,营,营销,的,的基,础,础,客户,细,细分,使,使差,异,异化,成,成为,可,可能,，,，使,提,提供,的,的产,品,品和,服,服务,更,更直,接,接的,针,针对,某,某一,特,特定,客,客户,群,群,为什,么,么划,分,分客,户,户群,？,？,产品/服,务,务,理解,客,客户,对,对产,品,品的,偏,偏好,，,，针,对,对客,户,户的,需,需求,提,提供,产,产品,渠,道,道,理解,客,客户,对,对销,售,售和,服,服务,要,要求,，,，有,针,针对,性,性地,设,设计,销,销售,渠,渠道,推,广,广,理解,客,客户,对,对市,场,场活,动,动的,反,反应,和,和接,受,受程,度,度，,针,针对,客,客户,群,群制,定,定推,广,广策,略,略,价格,制,制定,理解,客,客户,的,的价,格,格敏,感,感度,，,，针,对,对客,户,户的,需,需求,制,制定,价,价格,战,战略,12,分析,型,型CRM,的,的子,工,工作,流,流关,系,系图,本次,客,客户,细,细分,项,项目,重,重点,对,对客,户,户行,为,为及,需,需求,进,进行,分,分析,，,，及,对,对客,户,户细,分,分群,进,进行,定,定义,。,。,客户行为及需求分析,流失预测分析,流失原因分析,客户价值,客户细分群的定义,客户策略,挽留策略,客户生命周期价值,交义销售和增量销售,其他分析模型,客户价值,客户细分,客户周期价值,客户流失,其他分析模型,13,分析,型,型CRM,各,各模,型,型的,相,相互,关,关系,和,和用,处,处,模型,互,互相,作,作用-输,入,入方,向,向,客户,价,价值,模,模型,交叉,销,销售,和,和增,量,量,销,销售,发,发现,模,模型,应用,模,模型,流失,预,预测,模,模型,客户,细,细分,模,模型,1,2,识别,高,高离,网,网率,的,的高,价,价值,客,客户,预估,将,将来,最,最有,可,可能,离,离网,的,的客,户,户,可更,准,准确,地,地区,分,分高,价,价值,的,的客,户,户群,因,客,客户,价,价值,代,代表,每,每个,客,客户,群,群在,其,其生,命,命周,期,期内,为,为四,川,川移,动,动所,带,带来,的,的价,值,值,商业,应,应用,1,根据,客,客户,细,细分,模,模型,的,的,输,输出,物,物,再,再加,上,上每,个,个应,用,用模,型,型的,不,不同,分,分析,便,便能,把,把客,户,户群,分,分得,更,更准,确,确、,更,更清,晰,晰,1,1,2,3,客户,行,行为,、,、客,户,户价,值,值、,客,客户,需,需要,识别,客,客户,群,群的,交,交叉,和,和增,量,量销,售,售机,会,会,从,从而,增,增加,收,收入,了解,客,客户,对,对产,品,品/,服,服务,的,的需,求,求,支持,模,模型,客户,周,周期,价,价值,分析,模,模型,3,渠道,盈,盈利,分,分,析,析模,型,型,挽留,响,响应,模,模型,4,5,4,对各,种,种渠,道,道的,成,成本,所,所带,来,来的,收,收益,及,及付,出,出的,成,成本,有,有更,请,请楚,和,和准,确,确的,了,了解,对发,展,展客,户,户所,付,付出,的,的佣,金,金作,合,合理,水,水平,分,分析,以,方,方便,将,将来,作,作为,品,品牌,代,代理,的,的分,折,折,5,预估,客,客户,对,对客,户,户挽,留,留活,动,动有,响,响应,的,的可,能,能性,分析,现,现有,营,营销,活,活动,及,及集,中,中目,标,标、,减,减低,投,投资,分析,型,型CRM,通,通过,建,建立,模,模型,实,实现,知,知识,分,分析,平,平台,将,客,客户,数,数据,转,转化,为,为对,客,客户,的,的了,解,解,并,并由,此,此产,生,生有,针,针对,性,性的,运,运作,14,可供,选,选择,的,的客,户,户细,分,分方,法,法,具体,选,选用,哪,哪种,方,方法,需,需要,考,考量,实,实施,细,细分,方,方法,的,的难,易,易程,度,度及,有,有效,程,程度,，,，同,时,时还,要,要增,加,加股,东,东价,值,值和,满,满足,决,决策,目,目标,实施,的,的难,易,易程,度,度,人口,统,统计(Demographical),客户,价,价值,(Cu,stomerValue),行为,方,方式(Behavioral),态度(Attitudinal),性别,年龄,户藉,职业,婚姻状况,教育程度,收入,通话时段,繁忙和非繁忙通话量,漫游服务,方便程度,行为方式的变化,高利润率,中等利润率,低利润率,负利润率,形象,价值观,生活方式,心理因素,人口统计,行为方式,客户价值,态度,客户细分方式范例,15,各种,客,客户,细,细分,方,方法,的,的比,较,较,任何,客,客户,细,细分,都,都会,运,运用,到,到多,维,维的,客,客户,细,细分,方,方法,。,。每,个,个细,分,分方,法,法都,有,有其,优,优缺,点,点。,细分,基,基础,优势,劣势,应用,客户特征,人口统计,所处的生命周期,对营业厅的接近度,可计量的,可获得的 (容易得到外部数据),难以和使用行为联系起来,客户保持及获取战略,客户行为,使用频率,使用量/价值,手机型号偏好,使用产品,应用服务,能被多次记录及确认,在一定范围内可实行,已有内部客户数据,可将使用记录入数据库中,根据对客户行为的初步分析作出决定,不可前瞻,数据可信度,业务战略,战略性客户细分的选择,重点的财务分析,数据库可为公司各业务的营销战略提供支持,历史数据分析,客户态度,价格敏感度,对促销的敏感度,品牌忠诚度,整体满意度,满意的程度,可了解客户态度，提高有效性,可计量的,在一定范围内可实行,不可评估、确认,难以获得,难以和外部数据相连接,和内部评估没有直接联系,目标客户获取及保留,新产品开发,品牌战略,客户需求的确认,认识改善的机会,16,我们,在,在四,川,川移,动,动使,用,用的,客,客户,细,细分,方,方法,考虑,不,不同,客,客户,细,细分,方,方法,的,的有,效,效性,及,及实,施,施的,难,难易,程,程度,，,，并,结,结合,移,移动,所,所提,供,供的,数,数据,资,资源,和,和移,动,动用,户,户特,点,点，,我,我们,采,采用,用,用户,行,行为,方,方式,进,进行,客,客户,细,细分,，,，再,结,结合,人,人口,统,统计,和,和客,户,户价,值,值准,确,确定,位,位细,分,分人,群,群,。,客户,价,价值,人口统计,行为方式,客户群X,客户,细,细分,三,三维,分,分析,体,体系,三维分析体系所带来的业务利益,客户价值,识别高价值客户群，成为目标客户群,行为方式,如何针对目标客户群，根据客户行为分析，扩展及保留客户群，提供服务满足客户需求,人口统计,整理客户群的背景资料以便市场营销人员能找到目标客户群,17,报告,会,会议,程,程,一、,项,项目,背,背景,及,及前,阶,阶段,工,工作,回,回顾,二、,客,客户,细,细分,方,方法,论,论,三、,四,四川,移,移动,客,客户,细,细分,模,模型,四、,通,通用,客,客户,细,细分,结,结果,分,分析,及,及相,应,应市,场,场策,略,略,五、,短,短信,专,专题,客,客户,细,细分,结,结果,分,分析,及,及相,应,应市,场,场策,略,略,六、基,于,于客户,细,细分结,果,果的市,场,场营销,活,活动讨,论,论,18,本期客,户,户细分,模,模型,聚类,模,模型,聚类的,原,原理是,把,把具有,相,相近特,征,征的观,测,测值聚,集,集为一,组,组，保,证,证各组,间,间特征,的,的相异,性,性最大,，,，同组,内,内各观,测,测值特,征,征的相,似,似性最,大,大。,在本项,目,目中，,采,采用通,话,话行为,、,、数据,业,业务使,用,用情况,等,等作为,细,细分变,量,量，把,有,有相近,行,行为特,征,征的人,群,群聚为,一,一组。,各条记,录,录在细,分,分变量,空,空间的,透,透视图,点对点,短,短信,梦网短,信,信,本地通,话,话,通话行,为,为,数据业,务,务,长途通,话,话,各行为,特,特征在,空,空间的,位,位置相,对,对集中,，,，因此,被,被划分,为,为有一,定,定共同,行,行为特,征,征的客,户,户群,19,CRISP-DM（Cross-IndustryStandardProcess forData Mining）,具,具体,流,流程如,下,下图：,数据挖,掘,掘方法,论,论CRISP-DM,各个环,节,节顺序,进,进行，,但,但需要,不,不断地,循,循环往,复,复进行,数,数据探,索,索和模,型,型的调,优,优,20,各环节,所,所涉及,的,的内容,和,和人员,先不考,虑,虑循环,往,往复的,探,探索和,调,调优过,程,程，直,接,接顺序,考,考察各,个,个环节,确定商,业,业目标,ETL,建立模,型,型,数据收,集,集、管,理,理,数据探,索,索、修,改,改,各部门,访,访谈,计费中,心,心支持,数,数据,挖,挖掘工,程,程师,数据挖,掘,掘、商,业,业分析,、,、市场,营,营销人,员,员,商业理,解,解Business,Understanding,数据理,解,解,Data Understanding,数据准,备,备,DataPreparation,建立模,型,型Modeling,模型评,估,估,Evaluation,结果发,布,布,Deployment,模型调,优,优,应用策,略,略,CRISP-DM,21,步骤一,：,：商业,理,理解,我们基,于,于访谈,中,中了解,的,的客户,需,需求，,采,采用用,户,户行为,特,特征作,为,为细分,变,变量，,用,用户人,口,口统计,信,信息和,客,客户价,值,值作为,描,描述变,量,量，从,而,而定位,人,人群特,征,征。,对行为,特,特征从,以,以下几,个,个方面,来,来获取,信,信息,对多元化服务的需求程度,服务类型,对通话的多层次需求,本地、长途、漫游呼叫时长,本地、长途、漫游呼叫次数,呼叫时间、次数（繁忙/非繁忙时段，工作/休息时段）,呼叫类型（主叫、被叫、呼转）,对资费的敏感程度,IP使用情况,优惠时段通话情况,拨打1861次数,对方便性及信息实时性的需求,SMS使用次数,Monternet使用次数,WAP上网时间,GPRS数据流量,导出客户需求种类,客户数据,注：细,分,分变量,用,于,于进行,客,客户细,分,分的变,量,量。,描,描述,变,变量,将客,户,户细分,成,成各个,群,群体后,，,，各群,体,体的基,本,本特征,22,步骤一,：,：商业,理,理解,基于客,户,户需求,，,，并结,合,合以上,行,行为特,征,征选取,的,的方向,，,，我们,定,定义了,几,几组数,据,据作为,细,细分变,量,量,通话范围,本地通话,d_local,t_local,省内长途通话,d_toll_inprn,t_toll_inprn,省间长途通话,d_toll_btwprn,t_toll_btwprn,国际、港澳台长途通话,d_toll_htm+idd,t_toll_htm+idd,活动范围,省内漫游通话,d_rm_inprn,t_rm_inprn,省际漫游通话,d_rm_btwprn,t_rm_btwprn,国际漫游通话,d_rm_idd,t_rm_idd,IP使用情况,IP通话,d_ip,t_ip,与各运营商联系程度,网内通话比例,d_mob_vs_Ttl,t_mob_vs_Ttl,联通通话比例,d_uni_vs_Ttl,t_uni_vs_Ttl,小灵通通话比例,d_phs_vs_Ttl,t_phs_vs_Ttl,固话通话比例,d_fix_vs_Ttl,t_fix_vs_Ttl,呼转行为,d_fwd_total,t_fwd_total,数据业务使用情况,WAP使用,d_wap,t_wap,点对点及网间短信发送次数,t_sms_total,梦网短信发送次数,t_sms_mont,客 服,到营业厅次数,t_service,拨打1860次数,t_1860,拨打1861次数,t_1861,注：其,中,中d_X代表,时,时长，t_X,代,代表次,数,数,23,步骤一,：,：商业,理,理解,客户基本信息,age,special_fee,t_fwd_uni,male,sms_fee,d_fwd_fix,female,wap_fee,t_fwd_fix,GenderMissing,ip_fee,闲时忙时通话情况,d_pkon_total,CashPay,gprs_fee,t_pkon_total,PrePay,message_fee,d_pkoff_total,vip_mark,other_fee,t_pkoff_total,due_mark,主叫被叫情况,d_outgoing_total,工作时间、生活时间通话情况,d_worktime_total,tenure_in_days,t_outgoing_total,t_worktime_total,费用信息,ARPU,d_incoming_total,d_resttime_total,should_fee,t_incoming_total,t_resttime_total,favor_fee,联系人群范围,distinct_out_total,各项短信使用情况,t_sms_p2p,rent_fee,distinct_in_total,t_sms_link,local_fee,与联系紧密人群的主被叫情况,d_outgoing_topN,t_sms_topN,inprn_fee,t_outgoing_topN,sms_msg_length,btwprn_fee,d_incoming_topN,distinct_sms_send,htm_fee,t_incoming_topN,distinct_sp,idd_fee,各项呼转情况,d_fwd_mob,rm_inprn_fee,t_fwd_mob,rm_btwprn_fee,d_fwd_srv,rm_idd_fee,t_fwd_srv,rm_toll_fee,d_fwd_uni,通过细,分,分变量,将,将客户,进,进行细,分,分，再,通,通过以,下,下描述,变,变量定,位,位人群,，,，进一,步,步分析,人,人群特,征,征。,24,步骤二,：,：数据,理,理解,数,据,据源来,自,自经营,分,分析系,统,统,客户基,本,本信息,（,（DW_USR_DCUSTM）,用户帐务信,息,息（DW_FEE_SHOULDDM）,CDR信息,语音CDR,（,（DW_CALL_CDR）,IP业务CDR（DW_NEWBUSI_CDR）,短信业务CDR（DW_NEWBUSI_SMSCDR,）,）,梦网业务CDR（DW_NEWBUSI_MESGCDR）,客服信息,25,步骤三：数,据,据准备,宽表生成流,程,程,注：宽表是,将,将数据经过,组,组合所形成,的,的包括细分,变,变量和描述,变,变量的记录,表,表。,DSS目前,开放的表格,作为接口表,其他接口文,件,件,作为中间表,的,的,数据集1,作为中间表,的,的,数据集2,建模使用的,宽,宽表,数据集,建模使用的,宽,宽表,数据集,接口文件直,接,接参与宽表,的,的生成,26,步骤四：建,立,立模型,通过因子分,析,析找到变量,之,之间的关系,，,，并优化变,量,量组合。在,对,对模型结果,的,的分析中，,根,根据标准,群间差距,最,最大，群内,差,差距最小的,原,原则进行分,析,析，同时调,整,整变量组合,，,，以尽量接,近,近标准。以,此,此方式循环,，,，逐步使模,型,型得到优化,。,。,数据探索,建 摸,模型分析,因子分析,27,步骤四：建,立,立模型,数,数据探索,customer type,项目,百分比,数量,用户类型不详,0.01,7,B普通,73.44,95758,B重要客户A1,0.14,185,B重要客户A2,0.10,133,B重要客户A3,0.18,241,B重要客户A4,0.21,276,B集团客户B1,0.22,282,B集团客户B2,1.13,1473,B集团客户B3,2.25,2930,B集团客户B4,6.28,8193,B信誉客户C1,0.01,11,B信誉客户C2,0.03,40,B信誉客户C3,1.38,1793,B信誉客户C4,1.01,1321,B信誉客户C5,1.51,1968,B信誉客户C6,1.54,2006,B潜在大用户,10.29,13411,B老客户,0.05,70,B原重要,0.02,21,B公免,0.21,269,customer_status,项目,百分比,数量,正常,79.05,103066,冒高,17.43,22721,欠停,0.34,448,报停,1.69,2206,挂失,0.02,23,预销,1.28,1669,强开,0.05,62,强关,0.15,193,tenure_in_days,项目,百分比,数量,在网时间90天以上,0.79,102942,在网时间90天以内,0.21,27446,公免不能代,表,表普遍用户,行,行为，容易,对,对在聚类形,成,成噪音,在用户状态,中,中仅选择正,常,常用户,选择入网时,间,间90天以,上,上用户，保,证,证研究样本,拥,拥有完整的,研,研究期间数,据,据（2003年8-10月）,注：黄色部,分,分为去掉的,数,数据,28,步骤四：建,立,立模型,数,数据探索,在记录中发,现,现should_fee（用户应,收,收费）=favor_fee（优,惠,惠费）的用,户,户记录，详,细,细查看其记,录,录，比较异,常,常。在与乐,山,山移动分公,司,司访谈后了,解,解此类记录,多,多为用于测,试,试的号码。,于,于是去掉该,部,部分记录,customer_id,phone_no,should_fee,favor_fee,320035348,13708130041,381.76,381.76,320035349,13708130042,375.02,375.02,320035350,13708130043,797.27,797.27,320035351,13708130044,1090.68,1090.68,320035352,13708130045,6.66,6.66,320035353,13708130046,432.07,432.07,320035354,13708130047,1478.76,1478.76,部分should_fee=favor_fee的记录,29,步骤四：建,立,立模型,数,数据探索,短信发送人,数,数极多号码,有,有可能为移,动,动统一向用,户,户发送。,以distinct_sms_send（短,信,信发送的人,数,数）为例,在通过变量,的,的标记直接,去,去掉非正常,值,值外，还通,过,过查看数据,去,去掉极值，,以,以免极少数,用,用户的行为,对,对细分结果,形,形成噪音。,以d_local（本,地,地通话时长,）,）为例,极少数人的,个,个别行为可,能,能影响细分,群,群，所以在,研,研究其图形,分,分布后，去,掉,掉部分极值,30,乐山移动用,户,户的ARPU值分布,乐山的平均ARPU值,：,：156.48元,步骤四：建,立,立模型,数,数据探索,我们获得的,乐,乐山移动用,户,户记录共有130388条，经过,正,正常值选择,、,、极值处理,等,等，最后用,于,于研究的记,录,录有78339条。以,下,下数据探索,工,工作着重于,了,了解乐山移,动,动用户概貌,，,，以及发现,数,数据存在的,问,问题。,乐山移动用,户,户年龄分布,乐山移动用,户,户性别比例,乐山移动用,户,户区域分布,31,乐山移动用,户,户付费方式,步骤四：建,立,立模型,数,数据探索,乐山移动用,户,户付费信息,中,中，仅有现,金,金支票和预,付,付费用户两,种,种标记。建,议,议将信用卡,托,托收、各银,行,行代收等方,式,式进行标记,。,根据全球通,品,品牌整合计,划,划，市州地,区,区用户ARPU值在300元以上,可,可享有贵宾,卡,卡，即VIP用户。从,图,图中所示，,系,系统中VIP标记部分,有,有误。,乐山移动用,户,户VIP情,况,况,32,步骤四：建,立,立模型,生,生成客户细,分,分模型,使用SPSS数据挖掘,工,工具进行客,户,户细分,33,步骤五：模,型,型评估,以t_fix_vs_total为强势特征,的,的细分组t_fix_vs_total的分布,所有用户t_fix_vs_total（固话通话,比,比例）的分,布,布,细分前整个,用,用户群数据,落,落差较大。,细,细分后，各,组,组的强势变,量,量的分布趋,于,于平缓，可,看,看出，聚类,模,模型将具有,相,相似特征的,记,记录聚在一,组,组。,若在此步骤,中,中若发现模,型,型不够优化,，,，还需要回,到,到第四步骤,，,，甚至第三,步,步骤进行调,整,整,34,步骤六：结,果,果发布,对模型结果,进,进行分析，,并,并针对细分,群,群提出相应,策,策略，该部,分,分内容在下,一,一节详细讲,述,述,35,报告会议程,一、项目背,景,景及前阶段,工,工作回顾,二、客户细,分,分方法论,三、四川移,动,动客户细分,模,模型,四、通用客,户,户细分结果,分,分析及相应,市,市场策略,五、短信专,题,题客户细分,结,结果分析及,相,相应市场策,略,略,六、,基,基于,客,客户,细,细分,结,结果,的,的市,场,场营,销,销活,动,动讨,论,论,36,细分,结,结果,数,数据,表,表（12,个,个客,户,户细,分,分组,）,）,细,细分,变,变量,组1,组2,组3,组4,组5,组6,组7,组8,组9,组10,客户细分群,cluster-1,cluster-8,cluster-12,cluster-9,cluster-7,cluster-5,cluster-10,cluster-3,cluster-11,cluster-4,cluster-6,cluster-2,客户数,占总客户比例,0.065,0.090,0.197,0.193,0.102,0.158,0.125,0.011,0.035,0.020,0.005,0.001,ARPU,68.53,98.87,126.41,135.44,145.52,162.74,153.84,194.58,467.68,545.86,818.85,1422.02,本地通话时间,7874.74,9575.98,14219.88,15029.94,18305.44,20607.77,18176.30,15219.15,74130.90,47523.80,49344.00,39735.35,本地通话次数,80.93,103.79,161.61,170.09,211.23,238.01,200.97,140.33,822.74,497.04,521.28,406.72,省内长途通话时间,255.34,560.60,671.11,843.86,658.50,885.16,1159.61,987.36,3774.80,8389.12,8418.61,6134.88,省内长途通话次数,1.86,4.42,5.72,7.29,5.71,7.81,10.02,7.12,34.03,73.17,73.56,53.14,省际长途通话时间,110.32,305.75,293.05,409.41,269.26,373.12,533.24,485.09,1263.32,2209.19,15981.60,4346.05,省际长途通话次数,0.60,1.73,1.81,2.58,1.65,2.34,3.43,2.80,8.71,15.92,120.70,28.84,国际港澳台长途通话时间,0.32,6.78,1.63,2.82,1.00,0.89,2.56,0.22,10.42,7.91,31.48,4273.49,国际港澳台长途通话次数,0.00,0.04,0.01,0.02,0.01,0.01,0.02,0.00,0.06,0.05,0.22,37.47,省内漫游通话时间,338.40,591.40,781.29,1046.00,796.11,1108.62,1560.10,1362.62,4515.15,10459.79,9909.14,8736.74,省内漫游通话次数,3.24,5.70,7.91,10.64,8.17,11.53,15.91,12.84,46.96,102.90,94.05,83.40,省际漫游通话时间,34.12,191.22,169.02,253.75,124.42,180.73,311.17,257.00,538.17,1436.58,14768.83,4178.14,省际漫游通话次数,0.25,1.44,1.39,2.11,1.08,1.57,2.62,2.03,4.75,12.32,127.57,31.60,国际漫游通话时间,0.17,1.52,0.78,1.32,0.82,0.79,3.52,0.00,12.78,6.33,29.38,4633.49,国际漫游通话次数,0.00,0.01,0.01,0.01,0.01,0.01,0.04,0.00,0.11,0.05,0.24,43.30,IP,通话时间,492.19,600.45,584.37,688.80,599.59,729.46,1084.45,2591.80,1104.83,8322.41,6629.19,2273.72,IP,通话次数,2.35,3.26,3.44,3.97,3.40,4.31,6.08,10.48,6.63,53.23,40.87,11.07,呼叫转移时间,666.86,689.19,904.18,988.47,1190.04,1353.71,1233.80,807.35,3677.67,2933.65,3967.39,3909.30,呼叫转移次数,8.89,10.47,13.67,15.03,18.27,20.41,18.64,11.43,57.94,43.24,59.17,61.81,与移动通话时间,/,总通话时间,0.13,0.17,0.22,0.39,0.19,0.36,0.58,0.34,0.46,0.34,0.46,0.44,与移动通话次数,/,总通话次数,0.11,0.17,0.22,0.40,0.18,0.35,0.57,0.32,0.46,0.34,0.47,0.49,与联通通话时间,/,总通话时间,0.07,0.09,0.16,0.09,0.39,0.24,0.09,0.10,0.18,0.17,0.15,0.11,与联通通话次数,/,总通话次数,0.06,0.08,0.16,0.09,0.39,0.24,0.09,0.10,0.19,0.17,0.15,0.12,与小灵通通话时间,/,总通话时间,0.02,0.01,0.01,0.01,0.01,0.01,0.00,0.01,0.01,0.01,0.00,0.00,与小灵通通话次数,/,总通话次数,0.02,0.01,0.01,0.01,0.01,0.01,0.00,0.01,0.01,0.01,0.00,0.00,与固话通话时间,/,总通话时间,0.17,0.61,0.41,0.36,0.27,0.23,0.21,0.26,0.22,0.36,0.32,0.38,与固话通话次数,/,总通话次数,0.14,0.57,0.39,0.34,0.25,0.22,0.19,0.23,0.21,0.34,0.30,0.37,WAP,上网时间,19.12,6.46,8.84,9.14,8.96,10.44,15.70,92.46,26.46,16.09,2.78,128.37,WAP,上网次数,0.14,0.04,0.05,0.06,0.05,0.06,0.08,0.62,0.16,0.08,0.02,0.26,短信总次数,33.30,11.05,19.52,21.92,28.80,29.94,37.44,701.66,53.13,49.90,58.34,87.40,梦网短信次数,4.00,3.98,4.66,4.77,4.51,4.95,4.82,8.50,6.64,7.05,7.16,6.93,到营业厅次数,0.48,0.53,0.73,0.82,0.76,0.95,0.90,1.67,3.31,3.73,4.73,6.84,拨打,1860,次数,1.73,0.63,0.84,0.72,0.98,0.94,0.84,3.68,0.82,1.08,0.84,0.21,拨打,1861,次数,6.81,2.45,4.02,3.56,4.83,5.15,4.27,11.52,3.15,4.06,2.11,0.26,37,细分,结,结果,数,数据,表,表,描,述,述变,量,量（,一,一）,组1,组2,组3,组4,组5,组6,组7,组8,组9,组10,客户细分群,cluster-1,cluster-8,cluster-12,cluster-9,cluster-7,cluster-5,cluster-10,cluster-3,cluster-11,cluster-4,cluster-6,cluster-2,客户数,占总客户比例,0.065,0.090,0.197,0.193,0.102,0.158,0.125,0.011,0.035,0.020,0.005,0.001,ARPU,68.53,98.87,126.41,135.44,145.52,162.74,153.84,194.58,467.68,545.86,818.85,1422.02,年龄,30.99,35.39,34.19,34.32,32.79,32.92,32.74,24.18,33.79,33.72,34.93,35.95,男性比例,0.59,0.64,0.67,0.66,0.69,0.72,0.66,0.47,0.75,0.74,0.80,0.77,女性比例,0.29,0.27,0.25,0.25,0.22,0.20,0.25,0.44,0.16,0.19,0.15,0.12,性别不详比例,0.11,0.09,0.09,0.09,0.08,0.08,0.09,0.09,0.08,0.07,0.06,0.12,现金付费比例,0.09,0.19,0.19,0.21,0.16,0.18,0.18,0.08,0.51,0.51,0.59,0.86,预付费比例,0.91,0.81,0.81,0.79,0.84,0.82,0.82,0.92,0.49,0.49,0.41,0.14,VIP,比例,0.07,0.13,0.18,0.20,0.18,0.23,0.20,0.18,0.76,0.81,0.85,0.88,欠费比例,0.02,0.02,0.03,0.03,0.03,0.03,0.03,0.01,0.09,0.09,0.10,0.05,在网时长（天）,775.00,823.20,852.02,888.43,760.20,827.97,794.64,424.44,1214.86,1152.10,1192.93,1500.07,应收费,88.89,128.22,164.29,179.07,194.14,220.03,215.79,258.79,692.31,697.45,991.01,1552.63,优惠费,20.35,29.35,37.88,43.63,48.61,57.29,61.95,64.21,224.63,151.59,172.16,130.61,月租费,21.56,35.10,38.07,39.20,39.74,42.81,40.71,18.98,48.90,49.37,49.09,49.22,基本通话费,42.48,59.87,87.67,92.93,115.20,128.86,113.00,97.73,500.66,329.60,336.54,259.09,本地通话费比例,0.48,0.47,0.53,0.52,0.59,0.59,0.52,0.38,0.72,0.47,0.34,0.17,省内长途费,0.85,2.34,2.47,2.94,2.28,3.10,3.74,3.51,14.29,22.71,20.88,14.11,省际长途费,1.74,4.52,5.20,6.85,5.10,6.74,9.30,7.90,27.07,65.88,98.06,58.33,港澳台长途费,0.00,0.02,0.00,0.01,0.00,0.01,0.00,0.00,0.01,0.03,0.06,2.64,国际长途费,0.01,0.34,0.08,0.11,0.03,0.01,0.08,0.00,0.51,0.21,0.81,29.41,长途费比例,0.03,0.06,0.05,0.06,0.04,0.04,0.06,0.04,0.06,0.13,0.12,0.07,省内漫游费,3.48,6.71,8.74,11.67,8.67,12.08,17.24,14.85,48.02,127.08,110.90,97.69,省际漫游费,0.43,2.79,2.22,3.33,1.55,2.29,4.02,3.36,6.51,23.34,183.56,57.51,国际漫游费,0.02,0.22,0.12,0.26,0.14,0.13,0.50,0.00,2.20,0.94,6.27,706.27,漫游费比例,0.04,0.08,0.07,0.09,0.05,0.07,0.10,0.07,0.08,0.22,0.30,0.55,漫游长途费,0.24,1.89,1.39,2.05,0.90,1.35,2.34,1.79,4.74,12.04,116.55,202.86,特服费,6.63,6.75,7.34,7.58,7.74,8.40,8.15,6.17,13.79,14.36,19.12,17.94,SMS,费,4.59,1.38,2.11,2.24,2.40,2.63,3.71,81.09,3.38,4.32,4.42,6.14,WAP,费,0.03,0.01,0.01,0.01,0.02,0.02,0.02,0.12,0.04,0.03,0.00,0.28,IP,费,1.62,2.10,1.99,2.32,2.08,2.42,3.70,9.60,4.19,27.82,20.30,19.78,IP,费比例,0.57,0.23,0.22,0.19,0.25,0.22,0.24,0.73,0.09,0.28,0.09,0.06,GPRS,费,0.45,0.10,0.15,0.19,0.14,0.15,0.31,0.36,1.71,2.28,2.68,12.09,代收信息费,4.34,3.57,4.68,5.12,4.36,5.36,5.28,12.35,10.35,11.22,11.57,14.60,其它费,0.36,0.55,2.02,2.20,3.71,3.65,3.66,0.60,5.88,6.11,10.10,16.78,38,细分,结,结果,数,数据,表,表,描,述,述变,量,量（,二,二）,组1,组2,组3,组4,组5,组6,组7,组8,组9,组10,客户细分群,cluster-1,cluster-8,cluster-12,cluster-9,cluster-7,cluster-5,cluster-10,cluster-3,cluster-11,cluster-4,cluster-6,cluster-2,客户数,占总客户比例,0.065,0.090,0.197,0.193,0.102,0.158,0.125,0.011,0.035,0.020,0.005,0.001,ARPU,68.53,98.87,126.41,135.44,145.52,162.74,153.84,194.58,467.68,545.86,818.85,1422.02,主叫通话时间,5266.84,6145.73,7954.73,8442.79,10029.99,11172.89,10108.50,8461.86,38951.29,27426.91,34500.81,25890.70,主叫通话次数,51.54,64.18,89.51,94.98,114.78,128.15,112.57,82.95,426.79,279.85,336.41,252.40,被叫通话时间,2949.69,4299.05,7223.79,7831.84,9194.73,10677.88,9739.74,8211.85,40198.52,30501.98,38798.64,29502.33,被叫通话次数,32.29,46.56,80.56,85.92,104.77,121.02,102.79,68.33,439.92,306.02,376.67,285.09,主叫通话时长,/,被叫通话时长,1.79,1.43,1.10,1.08,1.09,1.05,1.04,1.03,0.97,0.90,0.89,0.88,主叫通话次数,/,被叫通话次数,1.60,1.38,1.11,1.11,1.10,1.06,1.10,1.21,0.97,0.91,0.89,0.89,总打出号码数,15.01,24.29,33.43,34.19,35.40,41.54,34.14,27.79,101.83,91.98,96.06,84.37,总打入号码数,11.94,19.57,32.24,32.83,33.09,40.78,32.26,26.12,118.52,99.59,113.95,106.14,打出号码,(,排前,10,名的,),的平均通话时长,471.38,405.00,455.52,489.91,597.25,627.64,642.80,672.67,1799.54,1484.61,1693.76,1107.05,打出号码,(,排前,10,名的,),的平均通话次数,3.57,3.42,4.38,4.76,6.12,6.49,6.28,5.02,19.40,13.33,15.16,10.30,打入号码,(,排前,10,名的,),的平均通话时长,222.44,274.29,374.48,416.74,529.11,562.86,583.99,498.81,1618.24,1200.39,1467.55,1035.47,打入号码,(,排前,10,名的,),的平均通话次数,2.00,2.39,3.63,4.01,5.45,5.84,5.48,3.33,17.83,11.78,13.96,11.12,TOP10,的主叫通话时长,/,被叫通话时长,2.12,1.48,1.22,1.18,1.13,1.12,1.10,1.35,1.11,1.24,1.15,1.07,TOP10,的主叫通话次数,/,被叫通话次数,1.79,1.43,1.21,1.19,1.12,1.11,1.15,1.51,1.09,1.13,1.09,0.93,呼