Permasalahan pengalaman kerja yang buruk sangat berpengaruh besar pada performa perusahaan salah satunya dapat menaikkan nilai indeks employee turnover. Perusahaan Tech Mahindra merupakan perusahaan multinasional yang bergerak di bidang konsultan TI dan memiliki lebih dari 120 ribu karyawan yang tersebar di seluruh dunia, dan saat ini sedang menghadapi masalah tingginya nilai employee turnover. Statistik menunjukkan dari tahun 2015 hingga tahun 2022, nilai turnover perusahaan secara global selalu cenderung di atas batas limit, dan ini menunjukkan masalah yang tidak kunjung selesai dan patut untuk diselidiki.Baik dan buruknya pengalaman kerja saat ini sering diulas oleh karyawan ataupun eks karyawan secara sukarela sebagai personal judgment melalui platform pengulas seperti Glassdoor, Indeed, dan AmbitionBox. Ulasan dari platform tersebut dapat ditambang datanya dan diolah lalu diambil intisarinya. Penelitian ini mencoba untuk mengadopsi teknologi analisis teks menggunakan machine learning seperti analisis sentimen dan pemodelan topik dalam mengetahui orientasi nyata ulasan beserta topik tersembunyi yang dibicarakan oleh pengulas. Algoritma klasifikasi Naïve Bayes Classifier, Support Vector Machine, Decision Tree, Random Forest, K-Nearest Neighbor, dan algoritma pemodelan topik Latent Dirichlet Allocation sukses diaplikasikan.Hasil evaluasi klasifikasi menunjukkan bahwa Random Forest memiliki performa terbaik dengan nilai akurasi keseluruhan sebesar 69.9%, dan penerapan pemodelan topik sukses menemukan 10 topik positif dan 9 topik negatif terkait pengalaman kerja di perusahaan Tech Mahindra. Berdasarkan hasil klasifikasi data juga ditemukan ulasan bebas memiliki populasi positif 66.54%, netral 14.49%, dan negatif 18.97%, dan setelah dianalisis lebih lanjut dengan teori Herzberg ditemukan bahwa working condition, growth, dan coworker relationship menjadi keunggulan dan kemudian salary, recognition, dan policies menjadi kekurangan perusahaan. Dengan fakta yang ditemukan melalui metode adopsi teknologi yang dibentuk, ini menunjukkan kemampuan analisis ulasan yang lebih robust dan akurat dan dapat digunakan untuk Tech Mahindra kedepannya dalam membentuk rekomendasi dalam menghadapi masalah kepuasan kerja.

---

The bad work experience problems have a big influence on the company performance, one of them can increase the value of employee turnover index. Tech Mahindra is a multinational company engaged in IT consulting and has more than 120 thousand employees spread all over the world, and is currently facing the problem of high employee turnover rates. Statistics show that from 2015 to 2022, the global company's turnover value has always been above the maximum limit, and this is a problem that has not been resolved and needs to be fixed as soon as possible.

Good and bad work experiences are often reviewed by employees or former employees voluntarily as personal judgment through many online review platforms like Glassdoor, Indeed, and AmbitionBox. Reviews data from the platform can be mined, processed and then extracted for the insight. This research tries to adopt text analysis technology using machine learning such as sentiment analysis and topic modeling in order to know the review orientation and hidden topics discussed by the reviewers. The classification algorithms such as Naïve Bayes Classifier, Support Vector Machine, Decision Tree, Random Forest, K-Nearest Neighbor, and Latent Dirichlet Allocation for topic modeling algorithms were successfully applied.

The results of the classification evaluation show that Random Forest has the best performance with total accuracy of 69.9%, furthermore in topic modeling, it successfully found 10 positive topics and 9 negative topics related to work experiences at Tech Mahindra. Based on the findings of the review classification, it was also found that free text review had a positive population of 66.54%, neutral 14.49%, and negative 18.97%, and after further analysis with Herzberg theory it was found that working condition, growth, and coworker relationship were the advantages of company and then salary, recognition, and policies were the company weakness. With the facts found through the adoption of technology that has been formed, it showing a more robust and accurate review analysis ability and can be used for Tech Mahindra in the future to gather recommendations in dealing with job satisfaction issues.