Fenomena perkembangan layanan perbankan semakin berubah seiring denganberkembangnya teknologi. Pada tahun 2014 tren layanan perbankan konvensionalbergeser pada layanan mesin ATM (Claudia, 2014), diikuti pergeseran pada tahun 2018dimana tren bergeser menjadi tren pembayaran elektronik secara global (World PaymentReport, 2018). Di sisi lain, pada tahun 2019 industri perbankan mendapat disrupsi dariperkembangan FinTech yang dalam operasional kesehariannya tidak membutuhkan asetdan biaya yang dibutuhkan oleh industri perbankan untuk operasional mesin ATM.Kemudian pergeseran secara signifikan terjadi pada tahun 2020, pandemi Covid-19memaksa behaviour pengguna mesin ATM-CRM berubah drastis. Dengan kondisi ini,penelitian ini bertujuan untuk mengevaluasi apakah mesin ATM-CRM masih memilikikemampuan untuk menghasilkan profit di masa depan (setelah pandemi), dengan kondisidata yang mengalami ketidaklengkapan (sparsity).Evaluasi profit dilakukan dengan melihat secara agregat karatekristik fitur transaksidan karakter site mesin ATM-CRM. Jenis transaksi yang dipakai dalam penelitian inimeliputi transaksi Pembayaran, Pembelian, dan Transfer dengan mengambil sampel diseluruh kantor Cabang dan Mall di wilayah Jabodetabek. Hubungan pengaruh antarfiturnya terhadap pembetukan profit sebelum dan sesudah pandemi dilihat menggunakanPCA (Principal Component Analysis) dan korelasi spearman. Kemudian dilakukanperamalan untuk melihat batas bawah (asymptomatic) kemampuan mesin ATM-CRMdalam menghasilkan profit menggunakan time series Holts Winter dan jaringan saraftiruan (Artificial Neural Network). Selanjutnya dilakukan clustering dimana ATM-CRMyang memiliki grafik time series yang sama di-cluster ke dalam 1 cluster yang samamenggunakan DTW (Dynamic Time Wrapping) Distance. Hasil clustering akanmenunjukkan karakter site yang mendukung terbentuknya profit ATM-CRM, yang dapatmembantu industri perbankan dalam decision making dalam pengembangan jaringanwilayah instalasi ATM-CRM atau pengembangan fitur layanan mesin ATM-CRM

---

The phenomenon of the development of banking services is changing along with the

development of technology. In 2014 the trend of conventional banking services shifted to

ATM machine services (Claudia, 2014), followed by a shift in 2018 where the trend

shifted to the trend of electronic payments globally (World Payment Report, 2018). On

the other hand, in 2019 the banking industry was disrupted by the development of

Financial Technology, which in its daily operations did not require the assets and costs

required by the banking industry to operate ATM machines. Then a significant shifting

occurred in 2020. The Covid-19 pandemic has forced behavior of ATM-CRM users to

change drastically. This condition has become the background of this study to evaluate

whether the ATM-CRM machine still has the ability to generate profits in the future (after

pandemic), in the condition where the data has sparsity.

Profit evaluation is done by analyzing characteristic of type of transaction and the site

of ATM-CRM aggregately. Sampling has been done upon ATM-CRM in all Branches and

Shopping Center at Jabodetabek area, by taking particularly 3 types of transactions;

payment, purchase, and transfer. The correlation amongst the transaction’s features are

being examined using PCA (Principal Component Analysis) and Sperman. Time series

Holts Winter and ANN are being used to predict profit ATM-CRM and see its

asymptomatic condition, and then ATM-CRM which has similarity on its time series’s

graphic are being clustered using DTW (Dynamic Time Wrapping) Distance. These 3

clusters will refer site character which can enhance ability ATM-CRM on generating

profit, and therefore, can be used as justification for banking industry either in developing

area for installation of ATM-CRM machine, or enhancing new feature to boost

transactions on ATM-CRM