Synthesizing highly crystalline carbon nanotubes (CNTs) is important, but mass production of highly crystalline CNTs still remains a challenge. There is a lack of research on synthesizing highly crystalline CNTs, especially regarding the effect of carbon precursors. Comparing the results of CNT synthesis from different carbon precursors is difficult because of the interaction effect between parameters. In this paper, we used Bayesian optimization (BO) to obtain the optimum results and compare the results of CNT synthesis from different carbon precursors. Furthermore, we tested the synergetic effect between ethylene and acetylene on the crystallinity of CNTs with BO.In Chapter 2, before optimizing the CNT synthesis, for establishing the highly robust CNT synthesis system, we investigated the effect of vacuum baking and mitigating temperature overshooting. First, the vacuum baking recipe did not make a significant difference in the CV value of CNT synthesis, because our CNT synthesis system is inherently robust even with a standard recipe. Second, we developed the multi-step recipe and successfully mitigated the temperature overshooting from 25 to 2 °C. However, the mitigating temperature overshooting did not have a significant effect on the variability of CNT synthesis. In Chapter 3, to test the effect of carbon precursor on the crystallinity of CNTs, we compared the IG/ID of CNTs synthesized from different carbon precursors in optimum conditions obtained by Bayesian optimization. We optimized three synthesis conditions and compared the maximum IG/ID of CNTs synthesized from ethylene and acetylene. As a result, there is no significant difference between the maximum IG/ID of CNTs from ethylene and acetylene. However, our research suggested novel approaches to compare the results of CNT synthesis from different carbon precursors by obtaining the optimum with Bayesian optimization. In Chapter 4, we tested the synergetic effect between ethylene and acetylene on the crystallinity of CNTs. We optimized four synthesis conditions with BO and compared the maximum IG/ID of CNTs when using ethylene and acetylene simultaneously versus using them individually. As a result, IG/ID of CNTs when using ethylene and acetylene simultaneously is generally similar to CNTs when using them individually. Furthermore, the results of t-test of multiple linear regression with interaction term failed to conclude that there is a synergetic effect between ethylene and acetylene on the IG/ID of CNTs. However, CNTs synthesized when using ethylene and acetylene simultaneously generally exhibit a high carbon yield and IG/ID ratio compared to when using them alone. Our results proposed the possibility of overcoming the trade-off between high efficiency and high crystallinity by using two or more carbon precursors. We believed that these findings would significantly contribute to the mass production of highly crystalline CNTs.