Optimizing the measurement of the signal strength for Higgs-boson production in the $H \to \tau\tau$ decay using multivariate techniques at $\sqrt{s} = 13\,\text{TeV}$ with the ATLAS detector

The Higgs boson was first observed in July 2012 by the ATLAS and CMS experiments based at CERN. In a combined measurement of the two collaborations the mass was determined to $m_H = (125.09 \pm 0.21 \text{(stat.)} \pm 0.11 \text{(sys.)})\,\text{GeV}$. Further measurements confirmed the consistency with Standard-Model predictions for the Higgs boson. The $H \to \tau\tau$ decay channel is the most sensitive decay channel to probe the Yukawa couplings of the Higgs boson. This makes it an interesting channel to analyze during the second data-taking period of the LHC which started in 2015. In this thesis a multivariate approach based on boosted decision trees is developed to increase the sensitivity with respect to the cut-based analysis of the $H \to \tau^+\tau^- \to \ell^+ \ell^- 4 \nu$ decay channel for a combined 2015 and 2016 dataset corresponding to an integrated luminosity of $\mathcal{L} = 36.1\,\text{fb}^{-1}$ recorded with the ATLAS detector in proton-proton collisions at a center-of-mass energy of $\sqrt{s} = 13\,\text{TeV}$. The BDT hyperparameters and observables used as input variables are optimized in a $k$-fold cross-validation approach. The expected sensitivity of the cut-based analysis is $0.83\,\sigma$. Using the multivariate approach it was possible to increase the sensitivity by $63\,\%$ to $1.35\,\sigma$. Additionally, the expected uncertainties on the measurement of the signal strength $\mu = \sigma_\text{obs} / \sigma_\text{SM}$ are reduced from $1 \pm 1.27$ in the cut-based analysis to $1 \pm 0.68$ in the multivariate analysis.